diff options
Diffstat (limited to 'media/libvpx')
98 files changed, 4953 insertions, 694 deletions
diff --git a/media/libvpx/arm_cpu_runtime_detection_code_on_openbsd.patch b/media/libvpx/arm_cpu_runtime_detection_code_on_openbsd.patch index 4788b3996a..6dc899e8ba 100644 --- a/media/libvpx/arm_cpu_runtime_detection_code_on_openbsd.patch +++ b/media/libvpx/arm_cpu_runtime_detection_code_on_openbsd.patch @@ -6,10 +6,10 @@ Bug 1888772 - Allow ARM CPU runtime detection code to build on OpenBSD diff --git a/vpx_ports/aarch64_cpudetect.c b/vpx_ports/aarch64_cpudetect.c --- a/vpx_ports/aarch64_cpudetect.c +++ b/vpx_ports/aarch64_cpudetect.c -@@ -10,30 +10,30 @@ - +@@ -11,30 +11,30 @@ #include "./vpx_config.h" - #include "arm_cpudetect.h" + #include "vpx_ports/arm.h" + #include "vpx_ports/arm_cpudetect.h" #if defined(__APPLE__) #include <sys/sysctl.h> diff --git a/media/libvpx/config/linux/arm64/vp9_rtcd.h b/media/libvpx/config/linux/arm64/vp9_rtcd.h index b7d828d446..738de4f9f4 100644 --- a/media/libvpx/config/linux/arm64/vp9_rtcd.h +++ b/media/libvpx/config/linux/arm64/vp9_rtcd.h @@ -35,13 +35,11 @@ extern "C" { int64_t vp9_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); int64_t vp9_block_error_neon(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); -int64_t vp9_block_error_sve(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); -RTCD_EXTERN int64_t (*vp9_block_error)(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); +#define vp9_block_error vp9_block_error_neon int64_t vp9_block_error_fp_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, int block_size); int64_t vp9_block_error_fp_neon(const tran_low_t *coeff, const tran_low_t *dqcoeff, int block_size); -int64_t vp9_block_error_fp_sve(const tran_low_t *coeff, const tran_low_t *dqcoeff, int block_size); -RTCD_EXTERN int64_t (*vp9_block_error_fp)(const tran_low_t *coeff, const tran_low_t *dqcoeff, int block_size); +#define vp9_block_error_fp vp9_block_error_fp_neon int vp9_diamond_search_sad_c(const struct macroblock *x, const struct search_site_config *cfg, struct mv *ref_mv, uint32_t start_mv_sad, struct mv *best_mv, int search_param, int sad_per_bit, int *num00, const struct vp9_sad_table *sad_fn_ptr, const struct mv *center_mv); int vp9_diamond_search_sad_neon(const struct macroblock *x, const struct search_site_config *cfg, struct mv *ref_mv, uint32_t start_mv_sad, struct mv *best_mv, int search_param, int sad_per_bit, int *num00, const struct vp9_sad_table *sad_fn_ptr, const struct mv *center_mv); @@ -98,10 +96,6 @@ static void setup_rtcd_internal(void) (void)flags; - vp9_block_error = vp9_block_error_neon; - if (flags & HAS_SVE) vp9_block_error = vp9_block_error_sve; - vp9_block_error_fp = vp9_block_error_fp_neon; - if (flags & HAS_SVE) vp9_block_error_fp = vp9_block_error_fp_sve; } #endif diff --git a/media/libvpx/config/linux/arm64/vpx_config.asm b/media/libvpx/config/linux/arm64/vpx_config.asm index c51a76b3f6..a5b2d4303e 100644 --- a/media/libvpx/config/linux/arm64/vpx_config.asm +++ b/media/libvpx/config/linux/arm64/vpx_config.asm @@ -12,8 +12,8 @@ .equ HAVE_NEON , 1 .equ HAVE_NEON_DOTPROD , 1 .equ HAVE_NEON_I8MM , 1 -.equ HAVE_SVE , 1 -.equ HAVE_SVE2 , 1 +.equ HAVE_SVE , 0 +.equ HAVE_SVE2 , 0 .equ HAVE_MIPS32 , 0 .equ HAVE_DSPR2 , 0 .equ HAVE_MSA , 0 diff --git a/media/libvpx/config/linux/arm64/vpx_config.c b/media/libvpx/config/linux/arm64/vpx_config.c index c0d714503f..5c4f1798e5 100644 --- a/media/libvpx/config/linux/arm64/vpx_config.c +++ b/media/libvpx/config/linux/arm64/vpx_config.c @@ -6,5 +6,5 @@ /* in the file PATENTS. All contributing project authors may */ /* be found in the AUTHORS file in the root of the source tree. */ #include "vpx/vpx_codec.h" -static const char* const cfg = "--target=arm64-linux-gcc --enable-external-build --disable-examples --disable-install-docs --disable-unit-tests --enable-multi-res-encoding --size-limit=8192x4608 --enable-pic --disable-avx512 --enable-realtime-only --log=/home/cm/Work/gecko-dev/media/libvpx/config/linux/arm64/config.log"; +static const char* const cfg = "--target=arm64-linux-gcc --enable-external-build --disable-examples --disable-install-docs --disable-unit-tests --enable-multi-res-encoding --size-limit=8192x4608 --enable-pic --disable-avx512 --enable-realtime-only --disable-sve --log=/home/cm/Work/gecko-dev/media/libvpx/config/linux/arm64/config.log"; const char *vpx_codec_build_config(void) {return cfg;} diff --git a/media/libvpx/config/linux/arm64/vpx_config.h b/media/libvpx/config/linux/arm64/vpx_config.h index 12251ee0c1..526796879f 100644 --- a/media/libvpx/config/linux/arm64/vpx_config.h +++ b/media/libvpx/config/linux/arm64/vpx_config.h @@ -21,8 +21,8 @@ #define HAVE_NEON 1 #define HAVE_NEON_DOTPROD 1 #define HAVE_NEON_I8MM 1 -#define HAVE_SVE 1 -#define HAVE_SVE2 1 +#define HAVE_SVE 0 +#define HAVE_SVE2 0 #define HAVE_MIPS32 0 #define HAVE_DSPR2 0 #define HAVE_MSA 0 diff --git a/media/libvpx/config/linux/arm64/vpx_dsp_rtcd.h b/media/libvpx/config/linux/arm64/vpx_dsp_rtcd.h index 2c31ee4ef9..5a9b05ca14 100644 --- a/media/libvpx/config/linux/arm64/vpx_dsp_rtcd.h +++ b/media/libvpx/config/linux/arm64/vpx_dsp_rtcd.h @@ -916,8 +916,7 @@ void vpx_subtract_block_neon(int rows, int cols, int16_t *diff_ptr, ptrdiff_t di uint64_t vpx_sum_squares_2d_i16_c(const int16_t *src, int stride, int size); uint64_t vpx_sum_squares_2d_i16_neon(const int16_t *src, int stride, int size); -uint64_t vpx_sum_squares_2d_i16_sve(const int16_t *src, int stride, int size); -RTCD_EXTERN uint64_t (*vpx_sum_squares_2d_i16)(const int16_t *src, int stride, int size); +#define vpx_sum_squares_2d_i16 vpx_sum_squares_2d_i16_neon void vpx_tm_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); void vpx_tm_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); @@ -1149,8 +1148,6 @@ static void setup_rtcd_internal(void) if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_64x64x4d = vpx_sad_skip_64x64x4d_neon_dotprod; vpx_sse = vpx_sse_neon; if (flags & HAS_NEON_DOTPROD) vpx_sse = vpx_sse_neon_dotprod; - vpx_sum_squares_2d_i16 = vpx_sum_squares_2d_i16_neon; - if (flags & HAS_SVE) vpx_sum_squares_2d_i16 = vpx_sum_squares_2d_i16_sve; vpx_variance16x16 = vpx_variance16x16_neon; if (flags & HAS_NEON_DOTPROD) vpx_variance16x16 = vpx_variance16x16_neon_dotprod; vpx_variance16x32 = vpx_variance16x32_neon; diff --git a/media/libvpx/config/mac/arm64/vp8_rtcd.h b/media/libvpx/config/mac/arm64/vp8_rtcd.h new file mode 100644 index 0000000000..c8826511ff --- /dev/null +++ b/media/libvpx/config/mac/arm64/vp8_rtcd.h @@ -0,0 +1,204 @@ +// This file is generated. Do not edit. +#ifndef VP8_RTCD_H_ +#define VP8_RTCD_H_ + +#ifdef RTCD_C +#define RTCD_EXTERN +#else +#define RTCD_EXTERN extern +#endif + +/* + * VP8 + */ + +struct blockd; +struct macroblockd; +struct loop_filter_info; + +/* Encoder forward decls */ +struct block; +struct macroblock; +struct variance_vtable; +union int_mv; +struct yv12_buffer_config; + +#ifdef __cplusplus +extern "C" { +#endif + +void vp8_bilinear_predict16x16_c(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +void vp8_bilinear_predict16x16_neon(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +#define vp8_bilinear_predict16x16 vp8_bilinear_predict16x16_neon + +void vp8_bilinear_predict4x4_c(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +void vp8_bilinear_predict4x4_neon(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +#define vp8_bilinear_predict4x4 vp8_bilinear_predict4x4_neon + +void vp8_bilinear_predict8x4_c(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +void vp8_bilinear_predict8x4_neon(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +#define vp8_bilinear_predict8x4 vp8_bilinear_predict8x4_neon + +void vp8_bilinear_predict8x8_c(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +void vp8_bilinear_predict8x8_neon(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +#define vp8_bilinear_predict8x8 vp8_bilinear_predict8x8_neon + +int vp8_block_error_c(short *coeff, short *dqcoeff); +#define vp8_block_error vp8_block_error_c + +void vp8_copy32xn_c(const unsigned char *src_ptr, int src_stride, unsigned char *dst_ptr, int dst_stride, int height); +#define vp8_copy32xn vp8_copy32xn_c + +void vp8_copy_mem16x16_c(unsigned char *src, int src_stride, unsigned char *dst, int dst_stride); +void vp8_copy_mem16x16_neon(unsigned char *src, int src_stride, unsigned char *dst, int dst_stride); +#define vp8_copy_mem16x16 vp8_copy_mem16x16_neon + +void vp8_copy_mem8x4_c(unsigned char *src, int src_stride, unsigned char *dst, int dst_stride); +void vp8_copy_mem8x4_neon(unsigned char *src, int src_stride, unsigned char *dst, int dst_stride); +#define vp8_copy_mem8x4 vp8_copy_mem8x4_neon + +void vp8_copy_mem8x8_c(unsigned char *src, int src_stride, unsigned char *dst, int dst_stride); +void vp8_copy_mem8x8_neon(unsigned char *src, int src_stride, unsigned char *dst, int dst_stride); +#define vp8_copy_mem8x8 vp8_copy_mem8x8_neon + +void vp8_dc_only_idct_add_c(short input_dc, unsigned char *pred_ptr, int pred_stride, unsigned char *dst_ptr, int dst_stride); +void vp8_dc_only_idct_add_neon(short input_dc, unsigned char *pred_ptr, int pred_stride, unsigned char *dst_ptr, int dst_stride); +#define vp8_dc_only_idct_add vp8_dc_only_idct_add_neon + +int vp8_denoiser_filter_c(unsigned char *mc_running_avg_y, int mc_avg_y_stride, unsigned char *running_avg_y, int avg_y_stride, unsigned char *sig, int sig_stride, unsigned int motion_magnitude, int increase_denoising); +int vp8_denoiser_filter_neon(unsigned char *mc_running_avg_y, int mc_avg_y_stride, unsigned char *running_avg_y, int avg_y_stride, unsigned char *sig, int sig_stride, unsigned int motion_magnitude, int increase_denoising); +#define vp8_denoiser_filter vp8_denoiser_filter_neon + +int vp8_denoiser_filter_uv_c(unsigned char *mc_running_avg, int mc_avg_stride, unsigned char *running_avg, int avg_stride, unsigned char *sig, int sig_stride, unsigned int motion_magnitude, int increase_denoising); +int vp8_denoiser_filter_uv_neon(unsigned char *mc_running_avg, int mc_avg_stride, unsigned char *running_avg, int avg_stride, unsigned char *sig, int sig_stride, unsigned int motion_magnitude, int increase_denoising); +#define vp8_denoiser_filter_uv vp8_denoiser_filter_uv_neon + +void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *dest, int stride); +void vp8_dequant_idct_add_neon(short *input, short *dq, unsigned char *dest, int stride); +#define vp8_dequant_idct_add vp8_dequant_idct_add_neon + +void vp8_dequant_idct_add_uv_block_c(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs); +void vp8_dequant_idct_add_uv_block_neon(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs); +#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_neon + +void vp8_dequant_idct_add_y_block_c(short *q, short *dq, unsigned char *dst, int stride, char *eobs); +void vp8_dequant_idct_add_y_block_neon(short *q, short *dq, unsigned char *dst, int stride, char *eobs); +#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_neon + +void vp8_dequantize_b_c(struct blockd*, short *DQC); +void vp8_dequantize_b_neon(struct blockd*, short *DQC); +#define vp8_dequantize_b vp8_dequantize_b_neon + +int vp8_diamond_search_sad_c(struct macroblock *x, struct block *b, struct blockd *d, union int_mv *ref_mv, union int_mv *best_mv, int search_param, int sad_per_bit, int *num00, struct variance_vtable *fn_ptr, int *mvcost[2], union int_mv *center_mv); +#define vp8_diamond_search_sad vp8_diamond_search_sad_c + +void vp8_fast_quantize_b_c(struct block *, struct blockd *); +void vp8_fast_quantize_b_neon(struct block *, struct blockd *); +#define vp8_fast_quantize_b vp8_fast_quantize_b_neon + +void vp8_loop_filter_bh_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi); +void vp8_loop_filter_bh_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi); +#define vp8_loop_filter_bh vp8_loop_filter_bh_neon + +void vp8_loop_filter_bv_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi); +void vp8_loop_filter_bv_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi); +#define vp8_loop_filter_bv vp8_loop_filter_bv_neon + +void vp8_loop_filter_mbh_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi); +void vp8_loop_filter_mbh_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi); +#define vp8_loop_filter_mbh vp8_loop_filter_mbh_neon + +void vp8_loop_filter_mbv_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi); +void vp8_loop_filter_mbv_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi); +#define vp8_loop_filter_mbv vp8_loop_filter_mbv_neon + +void vp8_loop_filter_bhs_c(unsigned char *y_ptr, int y_stride, const unsigned char *blimit); +void vp8_loop_filter_bhs_neon(unsigned char *y_ptr, int y_stride, const unsigned char *blimit); +#define vp8_loop_filter_simple_bh vp8_loop_filter_bhs_neon + +void vp8_loop_filter_bvs_c(unsigned char *y_ptr, int y_stride, const unsigned char *blimit); +void vp8_loop_filter_bvs_neon(unsigned char *y_ptr, int y_stride, const unsigned char *blimit); +#define vp8_loop_filter_simple_bv vp8_loop_filter_bvs_neon + +void vp8_loop_filter_simple_horizontal_edge_c(unsigned char *y_ptr, int y_stride, const unsigned char *blimit); +void vp8_loop_filter_mbhs_neon(unsigned char *y_ptr, int y_stride, const unsigned char *blimit); +#define vp8_loop_filter_simple_mbh vp8_loop_filter_mbhs_neon + +void vp8_loop_filter_simple_vertical_edge_c(unsigned char *y_ptr, int y_stride, const unsigned char *blimit); +void vp8_loop_filter_mbvs_neon(unsigned char *y_ptr, int y_stride, const unsigned char *blimit); +#define vp8_loop_filter_simple_mbv vp8_loop_filter_mbvs_neon + +int vp8_mbblock_error_c(struct macroblock *mb, int dc); +#define vp8_mbblock_error vp8_mbblock_error_c + +int vp8_mbuverror_c(struct macroblock *mb); +#define vp8_mbuverror vp8_mbuverror_c + +int vp8_refining_search_sad_c(struct macroblock *x, struct block *b, struct blockd *d, union int_mv *ref_mv, int error_per_bit, int search_range, struct variance_vtable *fn_ptr, int *mvcost[2], union int_mv *center_mv); +#define vp8_refining_search_sad vp8_refining_search_sad_c + +void vp8_regular_quantize_b_c(struct block *, struct blockd *); +#define vp8_regular_quantize_b vp8_regular_quantize_b_c + +void vp8_short_fdct4x4_c(short *input, short *output, int pitch); +void vp8_short_fdct4x4_neon(short *input, short *output, int pitch); +#define vp8_short_fdct4x4 vp8_short_fdct4x4_neon + +void vp8_short_fdct8x4_c(short *input, short *output, int pitch); +void vp8_short_fdct8x4_neon(short *input, short *output, int pitch); +#define vp8_short_fdct8x4 vp8_short_fdct8x4_neon + +void vp8_short_idct4x4llm_c(short *input, unsigned char *pred_ptr, int pred_stride, unsigned char *dst_ptr, int dst_stride); +void vp8_short_idct4x4llm_neon(short *input, unsigned char *pred_ptr, int pred_stride, unsigned char *dst_ptr, int dst_stride); +#define vp8_short_idct4x4llm vp8_short_idct4x4llm_neon + +void vp8_short_inv_walsh4x4_c(short *input, short *mb_dqcoeff); +void vp8_short_inv_walsh4x4_neon(short *input, short *mb_dqcoeff); +#define vp8_short_inv_walsh4x4 vp8_short_inv_walsh4x4_neon + +void vp8_short_inv_walsh4x4_1_c(short *input, short *mb_dqcoeff); +#define vp8_short_inv_walsh4x4_1 vp8_short_inv_walsh4x4_1_c + +void vp8_short_walsh4x4_c(short *input, short *output, int pitch); +void vp8_short_walsh4x4_neon(short *input, short *output, int pitch); +#define vp8_short_walsh4x4 vp8_short_walsh4x4_neon + +void vp8_sixtap_predict16x16_c(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +void vp8_sixtap_predict16x16_neon(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +#define vp8_sixtap_predict16x16 vp8_sixtap_predict16x16_neon + +void vp8_sixtap_predict4x4_c(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +void vp8_sixtap_predict4x4_neon(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +#define vp8_sixtap_predict4x4 vp8_sixtap_predict4x4_neon + +void vp8_sixtap_predict8x4_c(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +void vp8_sixtap_predict8x4_neon(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +#define vp8_sixtap_predict8x4 vp8_sixtap_predict8x4_neon + +void vp8_sixtap_predict8x8_c(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +void vp8_sixtap_predict8x8_neon(unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch); +#define vp8_sixtap_predict8x8 vp8_sixtap_predict8x8_neon + +void vp8_temporal_filter_apply_c(unsigned char *frame1, unsigned int stride, unsigned char *frame2, unsigned int block_size, int strength, int filter_weight, unsigned int *accumulator, unsigned short *count); +#define vp8_temporal_filter_apply vp8_temporal_filter_apply_c + +void vp8_rtcd(void); + +#include "vpx_config.h" + +#ifdef RTCD_C +#include "vpx_ports/arm.h" +static void setup_rtcd_internal(void) +{ + int flags = arm_cpu_caps(); + + (void)flags; + +} +#endif + +#ifdef __cplusplus +} // extern "C" +#endif + +#endif diff --git a/media/libvpx/config/mac/arm64/vp9_rtcd.h b/media/libvpx/config/mac/arm64/vp9_rtcd.h new file mode 100644 index 0000000000..b8fc782dfd --- /dev/null +++ b/media/libvpx/config/mac/arm64/vp9_rtcd.h @@ -0,0 +1,116 @@ +// This file is generated. Do not edit. +#ifndef VP9_RTCD_H_ +#define VP9_RTCD_H_ + +#ifdef RTCD_C +#define RTCD_EXTERN +#else +#define RTCD_EXTERN extern +#endif + +/* + * VP9 + */ + +#include "vpx/vpx_integer.h" +#include "vp9/common/vp9_common.h" +#include "vp9/common/vp9_enums.h" +#include "vp9/common/vp9_filter.h" + +struct macroblockd; + +/* Encoder forward decls */ +struct macroblock; +struct macroblock_plane; +struct vp9_sad_table; +struct ScanOrder; +struct search_site_config; +struct mv; +union int_mv; +struct yv12_buffer_config; + +#ifdef __cplusplus +extern "C" { +#endif + +void vp9_apply_temporal_filter_c(const uint8_t *y_src, int y_src_stride, const uint8_t *y_pre, int y_pre_stride, const uint8_t *u_src, const uint8_t *v_src, int uv_src_stride, const uint8_t *u_pre, const uint8_t *v_pre, int uv_pre_stride, unsigned int block_width, unsigned int block_height, int ss_x, int ss_y, int strength, const int *const blk_fw, int use_32x32, uint32_t *y_accumulator, uint16_t *y_count, uint32_t *u_accumulator, uint16_t *u_count, uint32_t *v_accumulator, uint16_t *v_count); +void vp9_apply_temporal_filter_neon(const uint8_t *y_src, int y_src_stride, const uint8_t *y_pre, int y_pre_stride, const uint8_t *u_src, const uint8_t *v_src, int uv_src_stride, const uint8_t *u_pre, const uint8_t *v_pre, int uv_pre_stride, unsigned int block_width, unsigned int block_height, int ss_x, int ss_y, int strength, const int *const blk_fw, int use_32x32, uint32_t *y_accumulator, uint16_t *y_count, uint32_t *u_accumulator, uint16_t *u_count, uint32_t *v_accumulator, uint16_t *v_count); +#define vp9_apply_temporal_filter vp9_apply_temporal_filter_neon + +int64_t vp9_block_error_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); +int64_t vp9_block_error_neon(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); +int64_t vp9_block_error_sve(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); +RTCD_EXTERN int64_t (*vp9_block_error)(const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz); + +int64_t vp9_block_error_fp_c(const tran_low_t *coeff, const tran_low_t *dqcoeff, int block_size); +int64_t vp9_block_error_fp_neon(const tran_low_t *coeff, const tran_low_t *dqcoeff, int block_size); +int64_t vp9_block_error_fp_sve(const tran_low_t *coeff, const tran_low_t *dqcoeff, int block_size); +RTCD_EXTERN int64_t (*vp9_block_error_fp)(const tran_low_t *coeff, const tran_low_t *dqcoeff, int block_size); + +int vp9_diamond_search_sad_c(const struct macroblock *x, const struct search_site_config *cfg, struct mv *ref_mv, uint32_t start_mv_sad, struct mv *best_mv, int search_param, int sad_per_bit, int *num00, const struct vp9_sad_table *sad_fn_ptr, const struct mv *center_mv); +int vp9_diamond_search_sad_neon(const struct macroblock *x, const struct search_site_config *cfg, struct mv *ref_mv, uint32_t start_mv_sad, struct mv *best_mv, int search_param, int sad_per_bit, int *num00, const struct vp9_sad_table *sad_fn_ptr, const struct mv *center_mv); +#define vp9_diamond_search_sad vp9_diamond_search_sad_neon + +void vp9_fht16x16_c(const int16_t *input, tran_low_t *output, int stride, int tx_type); +void vp9_fht16x16_neon(const int16_t *input, tran_low_t *output, int stride, int tx_type); +#define vp9_fht16x16 vp9_fht16x16_neon + +void vp9_fht4x4_c(const int16_t *input, tran_low_t *output, int stride, int tx_type); +void vp9_fht4x4_neon(const int16_t *input, tran_low_t *output, int stride, int tx_type); +#define vp9_fht4x4 vp9_fht4x4_neon + +void vp9_fht8x8_c(const int16_t *input, tran_low_t *output, int stride, int tx_type); +void vp9_fht8x8_neon(const int16_t *input, tran_low_t *output, int stride, int tx_type); +#define vp9_fht8x8 vp9_fht8x8_neon + +void vp9_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride); +#define vp9_fwht4x4 vp9_fwht4x4_c + +void vp9_iht16x16_256_add_c(const tran_low_t *input, uint8_t *dest, int stride, int tx_type); +void vp9_iht16x16_256_add_neon(const tran_low_t *input, uint8_t *dest, int stride, int tx_type); +#define vp9_iht16x16_256_add vp9_iht16x16_256_add_neon + +void vp9_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int stride, int tx_type); +void vp9_iht4x4_16_add_neon(const tran_low_t *input, uint8_t *dest, int stride, int tx_type); +#define vp9_iht4x4_16_add vp9_iht4x4_16_add_neon + +void vp9_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int stride, int tx_type); +void vp9_iht8x8_64_add_neon(const tran_low_t *input, uint8_t *dest, int stride, int tx_type); +#define vp9_iht8x8_64_add vp9_iht8x8_64_add_neon + +void vp9_quantize_fp_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, const struct macroblock_plane *const mb_plane, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const struct ScanOrder *const scan_order); +void vp9_quantize_fp_neon(const tran_low_t *coeff_ptr, intptr_t n_coeffs, const struct macroblock_plane *const mb_plane, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const struct ScanOrder *const scan_order); +#define vp9_quantize_fp vp9_quantize_fp_neon + +void vp9_quantize_fp_32x32_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, const struct macroblock_plane *const mb_plane, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const struct ScanOrder *const scan_order); +void vp9_quantize_fp_32x32_neon(const tran_low_t *coeff_ptr, intptr_t n_coeffs, const struct macroblock_plane *const mb_plane, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const struct ScanOrder *const scan_order); +#define vp9_quantize_fp_32x32 vp9_quantize_fp_32x32_neon + +void vp9_scale_and_extend_frame_c(const struct yv12_buffer_config *src, struct yv12_buffer_config *dst, INTERP_FILTER filter_type, int phase_scaler); +void vp9_scale_and_extend_frame_neon(const struct yv12_buffer_config *src, struct yv12_buffer_config *dst, INTERP_FILTER filter_type, int phase_scaler); +#define vp9_scale_and_extend_frame vp9_scale_and_extend_frame_neon + +void vp9_rtcd(void); + +#include "vpx_config.h" + +#ifdef RTCD_C +#include "vpx_ports/arm.h" +static void setup_rtcd_internal(void) +{ + int flags = arm_cpu_caps(); + + (void)flags; + + vp9_block_error = vp9_block_error_neon; + if (flags & HAS_SVE) vp9_block_error = vp9_block_error_sve; + vp9_block_error_fp = vp9_block_error_fp_neon; + if (flags & HAS_SVE) vp9_block_error_fp = vp9_block_error_fp_sve; +} +#endif + +#ifdef __cplusplus +} // extern "C" +#endif + +#endif diff --git a/media/libvpx/config/mac/arm64/vpx_config.asm b/media/libvpx/config/mac/arm64/vpx_config.asm new file mode 100644 index 0000000000..82a96985ec --- /dev/null +++ b/media/libvpx/config/mac/arm64/vpx_config.asm @@ -0,0 +1,98 @@ +@ This file was created from a .asm file +@ using the ads2gas.pl script. +.syntax unified +.equ VPX_ARCH_ARM , 1 +.equ VPX_ARCH_AARCH64 , 1 +.equ VPX_ARCH_MIPS , 0 +.equ VPX_ARCH_X86 , 0 +.equ VPX_ARCH_X86_64 , 0 +.equ VPX_ARCH_PPC , 0 +.equ VPX_ARCH_LOONGARCH , 0 +.equ HAVE_NEON_ASM , 0 +.equ HAVE_NEON , 1 +.equ HAVE_NEON_DOTPROD , 1 +.equ HAVE_NEON_I8MM , 1 +.equ HAVE_SVE , 1 +.equ HAVE_SVE2 , 1 +.equ HAVE_MIPS32 , 0 +.equ HAVE_DSPR2 , 0 +.equ HAVE_MSA , 0 +.equ HAVE_MIPS64 , 0 +.equ HAVE_MMX , 0 +.equ HAVE_SSE , 0 +.equ HAVE_SSE2 , 0 +.equ HAVE_SSE3 , 0 +.equ HAVE_SSSE3 , 0 +.equ HAVE_SSE4_1 , 0 +.equ HAVE_AVX , 0 +.equ HAVE_AVX2 , 0 +.equ HAVE_AVX512 , 0 +.equ HAVE_VSX , 0 +.equ HAVE_MMI , 0 +.equ HAVE_LSX , 0 +.equ HAVE_LASX , 0 +.equ HAVE_VPX_PORTS , 1 +.equ HAVE_PTHREAD_H , 1 +.equ CONFIG_DEPENDENCY_TRACKING , 1 +.equ CONFIG_EXTERNAL_BUILD , 1 +.equ CONFIG_INSTALL_DOCS , 0 +.equ CONFIG_INSTALL_BINS , 1 +.equ CONFIG_INSTALL_LIBS , 1 +.equ CONFIG_INSTALL_SRCS , 0 +.equ CONFIG_DEBUG , 0 +.equ CONFIG_GPROF , 0 +.equ CONFIG_GCOV , 0 +.equ CONFIG_RVCT , 0 +.equ CONFIG_GCC , 1 +.equ CONFIG_MSVS , 0 +.equ CONFIG_PIC , 1 +.equ CONFIG_BIG_ENDIAN , 0 +.equ CONFIG_CODEC_SRCS , 0 +.equ CONFIG_DEBUG_LIBS , 0 +.equ CONFIG_DEQUANT_TOKENS , 0 +.equ CONFIG_DC_RECON , 0 +.equ CONFIG_RUNTIME_CPU_DETECT , 1 +.equ CONFIG_POSTPROC , 0 +.equ CONFIG_VP9_POSTPROC , 0 +.equ CONFIG_MULTITHREAD , 1 +.equ CONFIG_INTERNAL_STATS , 0 +.equ CONFIG_VP8_ENCODER , 1 +.equ CONFIG_VP8_DECODER , 1 +.equ CONFIG_VP9_ENCODER , 1 +.equ CONFIG_VP9_DECODER , 1 +.equ CONFIG_VP8 , 1 +.equ CONFIG_VP9 , 1 +.equ CONFIG_ENCODERS , 1 +.equ CONFIG_DECODERS , 1 +.equ CONFIG_STATIC_MSVCRT , 0 +.equ CONFIG_SPATIAL_RESAMPLING , 1 +.equ CONFIG_REALTIME_ONLY , 0 +.equ CONFIG_ONTHEFLY_BITPACKING , 0 +.equ CONFIG_ERROR_CONCEALMENT , 0 +.equ CONFIG_SHARED , 0 +.equ CONFIG_STATIC , 1 +.equ CONFIG_SMALL , 0 +.equ CONFIG_POSTPROC_VISUALIZER , 0 +.equ CONFIG_OS_SUPPORT , 1 +.equ CONFIG_UNIT_TESTS , 0 +.equ CONFIG_WEBM_IO , 0 +.equ CONFIG_LIBYUV , 0 +.equ CONFIG_DECODE_PERF_TESTS , 0 +.equ CONFIG_ENCODE_PERF_TESTS , 0 +.equ CONFIG_MULTI_RES_ENCODING , 1 +.equ CONFIG_TEMPORAL_DENOISING , 1 +.equ CONFIG_VP9_TEMPORAL_DENOISING , 0 +.equ CONFIG_COEFFICIENT_RANGE_CHECKING , 0 +.equ CONFIG_VP9_HIGHBITDEPTH , 0 +.equ CONFIG_BETTER_HW_COMPATIBILITY , 0 +.equ CONFIG_EXPERIMENTAL , 0 +.equ CONFIG_SIZE_LIMIT , 1 +.equ CONFIG_ALWAYS_ADJUST_BPM , 0 +.equ CONFIG_BITSTREAM_DEBUG , 0 +.equ CONFIG_MISMATCH_DEBUG , 0 +.equ CONFIG_FP_MB_STATS , 0 +.equ CONFIG_EMULATE_HARDWARE , 0 +.equ CONFIG_NON_GREEDY_MV , 0 +.equ CONFIG_RATE_CTRL , 0 +.equ CONFIG_COLLECT_COMPONENT_TIMING , 0 + .section .note.GNU-stack,"",%progbits diff --git a/media/libvpx/config/mac/arm64/vpx_config.c b/media/libvpx/config/mac/arm64/vpx_config.c new file mode 100644 index 0000000000..3cf744dc16 --- /dev/null +++ b/media/libvpx/config/mac/arm64/vpx_config.c @@ -0,0 +1,10 @@ +/* Copyright (c) 2011 The WebM project authors. All Rights Reserved. */ +/* */ +/* Use of this source code is governed by a BSD-style license */ +/* that can be found in the LICENSE file in the root of the source */ +/* tree. An additional intellectual property rights grant can be found */ +/* in the file PATENTS. All contributing project authors may */ +/* be found in the AUTHORS file in the root of the source tree. */ +#include "vpx/vpx_codec.h" +static const char* const cfg = "--target=arm64-darwin-gcc --enable-external-build --disable-examples --disable-install-docs --disable-unit-tests --enable-multi-res-encoding --size-limit=8192x4608 --enable-pic --disable-avx512 --log=/home/cm/Work/gecko-dev/media/libvpx/config/mac/arm64/config.log"; +const char *vpx_codec_build_config(void) {return cfg;} diff --git a/media/libvpx/config/mac/arm64/vpx_config.h b/media/libvpx/config/mac/arm64/vpx_config.h new file mode 100644 index 0000000000..098ae3eff4 --- /dev/null +++ b/media/libvpx/config/mac/arm64/vpx_config.h @@ -0,0 +1,109 @@ +/* Copyright (c) 2011 The WebM project authors. All Rights Reserved. */ +/* */ +/* Use of this source code is governed by a BSD-style license */ +/* that can be found in the LICENSE file in the root of the source */ +/* tree. An additional intellectual property rights grant can be found */ +/* in the file PATENTS. All contributing project authors may */ +/* be found in the AUTHORS file in the root of the source tree. */ +/* This file automatically generated by configure. Do not edit! */ +#ifndef VPX_CONFIG_H +#define VPX_CONFIG_H +#define RESTRICT +#define INLINE inline +#define VPX_ARCH_ARM 1 +#define VPX_ARCH_AARCH64 1 +#define VPX_ARCH_MIPS 0 +#define VPX_ARCH_X86 0 +#define VPX_ARCH_X86_64 0 +#define VPX_ARCH_PPC 0 +#define VPX_ARCH_LOONGARCH 0 +#define HAVE_NEON_ASM 0 +#define HAVE_NEON 1 +#define HAVE_NEON_DOTPROD 1 +#define HAVE_NEON_I8MM 1 +#define HAVE_SVE 1 +#define HAVE_SVE2 1 +#define HAVE_MIPS32 0 +#define HAVE_DSPR2 0 +#define HAVE_MSA 0 +#define HAVE_MIPS64 0 +#define HAVE_MMX 0 +#define HAVE_SSE 0 +#define HAVE_SSE2 0 +#define HAVE_SSE3 0 +#define HAVE_SSSE3 0 +#define HAVE_SSE4_1 0 +#define HAVE_AVX 0 +#define HAVE_AVX2 0 +#define HAVE_AVX512 0 +#define HAVE_VSX 0 +#define HAVE_MMI 0 +#define HAVE_LSX 0 +#define HAVE_LASX 0 +#define HAVE_VPX_PORTS 1 +#define HAVE_PTHREAD_H 1 +#define CONFIG_DEPENDENCY_TRACKING 1 +#define CONFIG_EXTERNAL_BUILD 1 +#define CONFIG_INSTALL_DOCS 0 +#define CONFIG_INSTALL_BINS 1 +#define CONFIG_INSTALL_LIBS 1 +#define CONFIG_INSTALL_SRCS 0 +#define CONFIG_DEBUG 0 +#define CONFIG_GPROF 0 +#define CONFIG_GCOV 0 +#define CONFIG_RVCT 0 +#define CONFIG_GCC 1 +#define CONFIG_MSVS 0 +#define CONFIG_PIC 1 +#define CONFIG_BIG_ENDIAN 0 +#define CONFIG_CODEC_SRCS 0 +#define CONFIG_DEBUG_LIBS 0 +#define CONFIG_DEQUANT_TOKENS 0 +#define CONFIG_DC_RECON 0 +#define CONFIG_RUNTIME_CPU_DETECT 1 +#define CONFIG_POSTPROC 0 +#define CONFIG_VP9_POSTPROC 0 +#define CONFIG_MULTITHREAD 1 +#define CONFIG_INTERNAL_STATS 0 +#define CONFIG_VP8_ENCODER 1 +#define CONFIG_VP8_DECODER 1 +#define CONFIG_VP9_ENCODER 1 +#define CONFIG_VP9_DECODER 1 +#define CONFIG_VP8 1 +#define CONFIG_VP9 1 +#define CONFIG_ENCODERS 1 +#define CONFIG_DECODERS 1 +#define CONFIG_STATIC_MSVCRT 0 +#define CONFIG_SPATIAL_RESAMPLING 1 +#define CONFIG_REALTIME_ONLY 0 +#define CONFIG_ONTHEFLY_BITPACKING 0 +#define CONFIG_ERROR_CONCEALMENT 0 +#define CONFIG_SHARED 0 +#define CONFIG_STATIC 1 +#define CONFIG_SMALL 0 +#define CONFIG_POSTPROC_VISUALIZER 0 +#define CONFIG_OS_SUPPORT 1 +#define CONFIG_UNIT_TESTS 0 +#define CONFIG_WEBM_IO 0 +#define CONFIG_LIBYUV 0 +#define CONFIG_DECODE_PERF_TESTS 0 +#define CONFIG_ENCODE_PERF_TESTS 0 +#define CONFIG_MULTI_RES_ENCODING 1 +#define CONFIG_TEMPORAL_DENOISING 1 +#define CONFIG_VP9_TEMPORAL_DENOISING 0 +#define CONFIG_COEFFICIENT_RANGE_CHECKING 0 +#define CONFIG_VP9_HIGHBITDEPTH 0 +#define CONFIG_BETTER_HW_COMPATIBILITY 0 +#define CONFIG_EXPERIMENTAL 0 +#define CONFIG_SIZE_LIMIT 1 +#define CONFIG_ALWAYS_ADJUST_BPM 0 +#define CONFIG_BITSTREAM_DEBUG 0 +#define CONFIG_MISMATCH_DEBUG 0 +#define CONFIG_FP_MB_STATS 0 +#define CONFIG_EMULATE_HARDWARE 0 +#define CONFIG_NON_GREEDY_MV 0 +#define CONFIG_RATE_CTRL 0 +#define CONFIG_COLLECT_COMPONENT_TIMING 0 +#define DECODE_WIDTH_LIMIT 8192 +#define DECODE_HEIGHT_LIMIT 4608 +#endif /* VPX_CONFIG_H */ diff --git a/media/libvpx/config/mac/arm64/vpx_dsp_rtcd.h b/media/libvpx/config/mac/arm64/vpx_dsp_rtcd.h new file mode 100644 index 0000000000..2c31ee4ef9 --- /dev/null +++ b/media/libvpx/config/mac/arm64/vpx_dsp_rtcd.h @@ -0,0 +1,1187 @@ +// This file is generated. Do not edit. +#ifndef VPX_DSP_RTCD_H_ +#define VPX_DSP_RTCD_H_ + +#ifdef RTCD_C +#define RTCD_EXTERN +#else +#define RTCD_EXTERN extern +#endif + +/* + * DSP + */ + +#include "vpx/vpx_integer.h" +#include "vpx_dsp/vpx_dsp_common.h" +#include "vpx_dsp/vpx_filter.h" +#if CONFIG_VP9_ENCODER + struct macroblock_plane; + struct ScanOrder; +#endif + + +#ifdef __cplusplus +extern "C" { +#endif + +unsigned int vpx_avg_4x4_c(const uint8_t *, int p); +unsigned int vpx_avg_4x4_neon(const uint8_t *, int p); +#define vpx_avg_4x4 vpx_avg_4x4_neon + +unsigned int vpx_avg_8x8_c(const uint8_t *, int p); +unsigned int vpx_avg_8x8_neon(const uint8_t *, int p); +#define vpx_avg_8x8 vpx_avg_8x8_neon + +void vpx_comp_avg_pred_c(uint8_t *comp_pred, const uint8_t *pred, int width, int height, const uint8_t *ref, int ref_stride); +void vpx_comp_avg_pred_neon(uint8_t *comp_pred, const uint8_t *pred, int width, int height, const uint8_t *ref, int ref_stride); +#define vpx_comp_avg_pred vpx_comp_avg_pred_neon + +void vpx_convolve8_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_neon_dotprod(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_neon_i8mm(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +RTCD_EXTERN void (*vpx_convolve8)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); + +void vpx_convolve8_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_avg_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_avg_neon_dotprod(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_avg_neon_i8mm(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +RTCD_EXTERN void (*vpx_convolve8_avg)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); + +void vpx_convolve8_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_avg_horiz_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_avg_horiz_neon_dotprod(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_avg_horiz_neon_i8mm(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +RTCD_EXTERN void (*vpx_convolve8_avg_horiz)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); + +void vpx_convolve8_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_avg_vert_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_avg_vert_neon_dotprod(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_avg_vert_neon_i8mm(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +RTCD_EXTERN void (*vpx_convolve8_avg_vert)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); + +void vpx_convolve8_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_horiz_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_horiz_neon_dotprod(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_horiz_neon_i8mm(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +RTCD_EXTERN void (*vpx_convolve8_horiz)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); + +void vpx_convolve8_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_vert_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_vert_neon_dotprod(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve8_vert_neon_i8mm(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +RTCD_EXTERN void (*vpx_convolve8_vert)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); + +void vpx_convolve_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve_avg_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +#define vpx_convolve_avg vpx_convolve_avg_neon + +void vpx_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_convolve_copy_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +#define vpx_convolve_copy vpx_convolve_copy_neon + +void vpx_d117_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d117_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d117_predictor_16x16 vpx_d117_predictor_16x16_neon + +void vpx_d117_predictor_32x32_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d117_predictor_32x32_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d117_predictor_32x32 vpx_d117_predictor_32x32_neon + +void vpx_d117_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d117_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d117_predictor_4x4 vpx_d117_predictor_4x4_neon + +void vpx_d117_predictor_8x8_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d117_predictor_8x8_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d117_predictor_8x8 vpx_d117_predictor_8x8_neon + +void vpx_d135_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d135_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d135_predictor_16x16 vpx_d135_predictor_16x16_neon + +void vpx_d135_predictor_32x32_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d135_predictor_32x32_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d135_predictor_32x32 vpx_d135_predictor_32x32_neon + +void vpx_d135_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d135_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d135_predictor_4x4 vpx_d135_predictor_4x4_neon + +void vpx_d135_predictor_8x8_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d135_predictor_8x8_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d135_predictor_8x8 vpx_d135_predictor_8x8_neon + +void vpx_d153_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d153_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d153_predictor_16x16 vpx_d153_predictor_16x16_neon + +void vpx_d153_predictor_32x32_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d153_predictor_32x32_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d153_predictor_32x32 vpx_d153_predictor_32x32_neon + +void vpx_d153_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d153_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d153_predictor_4x4 vpx_d153_predictor_4x4_neon + +void vpx_d153_predictor_8x8_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d153_predictor_8x8_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d153_predictor_8x8 vpx_d153_predictor_8x8_neon + +void vpx_d207_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d207_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d207_predictor_16x16 vpx_d207_predictor_16x16_neon + +void vpx_d207_predictor_32x32_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d207_predictor_32x32_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d207_predictor_32x32 vpx_d207_predictor_32x32_neon + +void vpx_d207_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d207_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d207_predictor_4x4 vpx_d207_predictor_4x4_neon + +void vpx_d207_predictor_8x8_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d207_predictor_8x8_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d207_predictor_8x8 vpx_d207_predictor_8x8_neon + +void vpx_d45_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d45_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d45_predictor_16x16 vpx_d45_predictor_16x16_neon + +void vpx_d45_predictor_32x32_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d45_predictor_32x32_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d45_predictor_32x32 vpx_d45_predictor_32x32_neon + +void vpx_d45_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d45_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d45_predictor_4x4 vpx_d45_predictor_4x4_neon + +void vpx_d45_predictor_8x8_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d45_predictor_8x8_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d45_predictor_8x8 vpx_d45_predictor_8x8_neon + +void vpx_d45e_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d45e_predictor_4x4 vpx_d45e_predictor_4x4_c + +void vpx_d63_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d63_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d63_predictor_16x16 vpx_d63_predictor_16x16_neon + +void vpx_d63_predictor_32x32_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d63_predictor_32x32_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d63_predictor_32x32 vpx_d63_predictor_32x32_neon + +void vpx_d63_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d63_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d63_predictor_4x4 vpx_d63_predictor_4x4_neon + +void vpx_d63_predictor_8x8_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_d63_predictor_8x8_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d63_predictor_8x8 vpx_d63_predictor_8x8_neon + +void vpx_d63e_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_d63e_predictor_4x4 vpx_d63e_predictor_4x4_c + +void vpx_dc_128_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_128_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_128_predictor_16x16 vpx_dc_128_predictor_16x16_neon + +void vpx_dc_128_predictor_32x32_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_128_predictor_32x32_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_128_predictor_32x32 vpx_dc_128_predictor_32x32_neon + +void vpx_dc_128_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_128_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_128_predictor_4x4 vpx_dc_128_predictor_4x4_neon + +void vpx_dc_128_predictor_8x8_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_128_predictor_8x8_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_128_predictor_8x8 vpx_dc_128_predictor_8x8_neon + +void vpx_dc_left_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_left_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_left_predictor_16x16 vpx_dc_left_predictor_16x16_neon + +void vpx_dc_left_predictor_32x32_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_left_predictor_32x32_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_left_predictor_32x32 vpx_dc_left_predictor_32x32_neon + +void vpx_dc_left_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_left_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_left_predictor_4x4 vpx_dc_left_predictor_4x4_neon + +void vpx_dc_left_predictor_8x8_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_left_predictor_8x8_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_left_predictor_8x8 vpx_dc_left_predictor_8x8_neon + +void vpx_dc_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_predictor_16x16 vpx_dc_predictor_16x16_neon + +void vpx_dc_predictor_32x32_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_predictor_32x32_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_predictor_32x32 vpx_dc_predictor_32x32_neon + +void vpx_dc_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_predictor_4x4 vpx_dc_predictor_4x4_neon + +void vpx_dc_predictor_8x8_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_predictor_8x8_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_predictor_8x8 vpx_dc_predictor_8x8_neon + +void vpx_dc_top_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_top_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_top_predictor_16x16 vpx_dc_top_predictor_16x16_neon + +void vpx_dc_top_predictor_32x32_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_top_predictor_32x32_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_top_predictor_32x32 vpx_dc_top_predictor_32x32_neon + +void vpx_dc_top_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_top_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_top_predictor_4x4 vpx_dc_top_predictor_4x4_neon + +void vpx_dc_top_predictor_8x8_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_dc_top_predictor_8x8_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_dc_top_predictor_8x8 vpx_dc_top_predictor_8x8_neon + +void vpx_fdct16x16_c(const int16_t *input, tran_low_t *output, int stride); +void vpx_fdct16x16_neon(const int16_t *input, tran_low_t *output, int stride); +#define vpx_fdct16x16 vpx_fdct16x16_neon + +void vpx_fdct16x16_1_c(const int16_t *input, tran_low_t *output, int stride); +void vpx_fdct16x16_1_neon(const int16_t *input, tran_low_t *output, int stride); +#define vpx_fdct16x16_1 vpx_fdct16x16_1_neon + +void vpx_fdct32x32_c(const int16_t *input, tran_low_t *output, int stride); +void vpx_fdct32x32_neon(const int16_t *input, tran_low_t *output, int stride); +#define vpx_fdct32x32 vpx_fdct32x32_neon + +void vpx_fdct32x32_1_c(const int16_t *input, tran_low_t *output, int stride); +void vpx_fdct32x32_1_neon(const int16_t *input, tran_low_t *output, int stride); +#define vpx_fdct32x32_1 vpx_fdct32x32_1_neon + +void vpx_fdct32x32_rd_c(const int16_t *input, tran_low_t *output, int stride); +void vpx_fdct32x32_rd_neon(const int16_t *input, tran_low_t *output, int stride); +#define vpx_fdct32x32_rd vpx_fdct32x32_rd_neon + +void vpx_fdct4x4_c(const int16_t *input, tran_low_t *output, int stride); +void vpx_fdct4x4_neon(const int16_t *input, tran_low_t *output, int stride); +#define vpx_fdct4x4 vpx_fdct4x4_neon + +void vpx_fdct4x4_1_c(const int16_t *input, tran_low_t *output, int stride); +void vpx_fdct4x4_1_neon(const int16_t *input, tran_low_t *output, int stride); +#define vpx_fdct4x4_1 vpx_fdct4x4_1_neon + +void vpx_fdct8x8_c(const int16_t *input, tran_low_t *output, int stride); +void vpx_fdct8x8_neon(const int16_t *input, tran_low_t *output, int stride); +#define vpx_fdct8x8 vpx_fdct8x8_neon + +void vpx_fdct8x8_1_c(const int16_t *input, tran_low_t *output, int stride); +void vpx_fdct8x8_1_neon(const int16_t *input, tran_low_t *output, int stride); +#define vpx_fdct8x8_1 vpx_fdct8x8_1_neon + +void vpx_get16x16var_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum); +void vpx_get16x16var_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum); +void vpx_get16x16var_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum); +RTCD_EXTERN void (*vpx_get16x16var)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum); + +unsigned int vpx_get4x4sse_cs_c(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride); +unsigned int vpx_get4x4sse_cs_neon(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride); +unsigned int vpx_get4x4sse_cs_neon_dotprod(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_get4x4sse_cs)(const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride); + +void vpx_get8x8var_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum); +void vpx_get8x8var_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum); +void vpx_get8x8var_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum); +RTCD_EXTERN void (*vpx_get8x8var)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum); + +unsigned int vpx_get_mb_ss_c(const int16_t *); +#define vpx_get_mb_ss vpx_get_mb_ss_c + +void vpx_h_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_h_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_h_predictor_16x16 vpx_h_predictor_16x16_neon + +void vpx_h_predictor_32x32_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_h_predictor_32x32_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_h_predictor_32x32 vpx_h_predictor_32x32_neon + +void vpx_h_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_h_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_h_predictor_4x4 vpx_h_predictor_4x4_neon + +void vpx_h_predictor_8x8_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_h_predictor_8x8_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_h_predictor_8x8 vpx_h_predictor_8x8_neon + +void vpx_hadamard_16x16_c(const int16_t *src_diff, ptrdiff_t src_stride, int16_t *coeff); +void vpx_hadamard_16x16_neon(const int16_t *src_diff, ptrdiff_t src_stride, int16_t *coeff); +#define vpx_hadamard_16x16 vpx_hadamard_16x16_neon + +void vpx_hadamard_32x32_c(const int16_t *src_diff, ptrdiff_t src_stride, int16_t *coeff); +void vpx_hadamard_32x32_neon(const int16_t *src_diff, ptrdiff_t src_stride, int16_t *coeff); +#define vpx_hadamard_32x32 vpx_hadamard_32x32_neon + +void vpx_hadamard_8x8_c(const int16_t *src_diff, ptrdiff_t src_stride, int16_t *coeff); +void vpx_hadamard_8x8_neon(const int16_t *src_diff, ptrdiff_t src_stride, int16_t *coeff); +#define vpx_hadamard_8x8 vpx_hadamard_8x8_neon + +void vpx_he_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_he_predictor_4x4 vpx_he_predictor_4x4_c + +void vpx_idct16x16_10_add_c(const tran_low_t *input, uint8_t *dest, int stride); +void vpx_idct16x16_10_add_neon(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_idct16x16_10_add vpx_idct16x16_10_add_neon + +void vpx_idct16x16_1_add_c(const tran_low_t *input, uint8_t *dest, int stride); +void vpx_idct16x16_1_add_neon(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_idct16x16_1_add vpx_idct16x16_1_add_neon + +void vpx_idct16x16_256_add_c(const tran_low_t *input, uint8_t *dest, int stride); +void vpx_idct16x16_256_add_neon(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_idct16x16_256_add vpx_idct16x16_256_add_neon + +void vpx_idct16x16_38_add_c(const tran_low_t *input, uint8_t *dest, int stride); +void vpx_idct16x16_38_add_neon(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_idct16x16_38_add vpx_idct16x16_38_add_neon + +void vpx_idct32x32_1024_add_c(const tran_low_t *input, uint8_t *dest, int stride); +void vpx_idct32x32_1024_add_neon(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_idct32x32_1024_add vpx_idct32x32_1024_add_neon + +void vpx_idct32x32_135_add_c(const tran_low_t *input, uint8_t *dest, int stride); +void vpx_idct32x32_135_add_neon(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_idct32x32_135_add vpx_idct32x32_135_add_neon + +void vpx_idct32x32_1_add_c(const tran_low_t *input, uint8_t *dest, int stride); +void vpx_idct32x32_1_add_neon(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_idct32x32_1_add vpx_idct32x32_1_add_neon + +void vpx_idct32x32_34_add_c(const tran_low_t *input, uint8_t *dest, int stride); +void vpx_idct32x32_34_add_neon(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_idct32x32_34_add vpx_idct32x32_34_add_neon + +void vpx_idct4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int stride); +void vpx_idct4x4_16_add_neon(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_idct4x4_16_add vpx_idct4x4_16_add_neon + +void vpx_idct4x4_1_add_c(const tran_low_t *input, uint8_t *dest, int stride); +void vpx_idct4x4_1_add_neon(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_idct4x4_1_add vpx_idct4x4_1_add_neon + +void vpx_idct8x8_12_add_c(const tran_low_t *input, uint8_t *dest, int stride); +void vpx_idct8x8_12_add_neon(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_idct8x8_12_add vpx_idct8x8_12_add_neon + +void vpx_idct8x8_1_add_c(const tran_low_t *input, uint8_t *dest, int stride); +void vpx_idct8x8_1_add_neon(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_idct8x8_1_add vpx_idct8x8_1_add_neon + +void vpx_idct8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int stride); +void vpx_idct8x8_64_add_neon(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_idct8x8_64_add vpx_idct8x8_64_add_neon + +int16_t vpx_int_pro_col_c(const uint8_t *ref, const int width); +int16_t vpx_int_pro_col_neon(const uint8_t *ref, const int width); +#define vpx_int_pro_col vpx_int_pro_col_neon + +void vpx_int_pro_row_c(int16_t hbuf[16], const uint8_t *ref, const int ref_stride, const int height); +void vpx_int_pro_row_neon(int16_t hbuf[16], const uint8_t *ref, const int ref_stride, const int height); +#define vpx_int_pro_row vpx_int_pro_row_neon + +void vpx_iwht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_iwht4x4_16_add vpx_iwht4x4_16_add_c + +void vpx_iwht4x4_1_add_c(const tran_low_t *input, uint8_t *dest, int stride); +#define vpx_iwht4x4_1_add vpx_iwht4x4_1_add_c + +void vpx_lpf_horizontal_16_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +void vpx_lpf_horizontal_16_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +#define vpx_lpf_horizontal_16 vpx_lpf_horizontal_16_neon + +void vpx_lpf_horizontal_16_dual_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +void vpx_lpf_horizontal_16_dual_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +#define vpx_lpf_horizontal_16_dual vpx_lpf_horizontal_16_dual_neon + +void vpx_lpf_horizontal_4_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +void vpx_lpf_horizontal_4_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +#define vpx_lpf_horizontal_4 vpx_lpf_horizontal_4_neon + +void vpx_lpf_horizontal_4_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1); +void vpx_lpf_horizontal_4_dual_neon(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1); +#define vpx_lpf_horizontal_4_dual vpx_lpf_horizontal_4_dual_neon + +void vpx_lpf_horizontal_8_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +void vpx_lpf_horizontal_8_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +#define vpx_lpf_horizontal_8 vpx_lpf_horizontal_8_neon + +void vpx_lpf_horizontal_8_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1); +void vpx_lpf_horizontal_8_dual_neon(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1); +#define vpx_lpf_horizontal_8_dual vpx_lpf_horizontal_8_dual_neon + +void vpx_lpf_vertical_16_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +void vpx_lpf_vertical_16_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +#define vpx_lpf_vertical_16 vpx_lpf_vertical_16_neon + +void vpx_lpf_vertical_16_dual_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +void vpx_lpf_vertical_16_dual_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +#define vpx_lpf_vertical_16_dual vpx_lpf_vertical_16_dual_neon + +void vpx_lpf_vertical_4_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +void vpx_lpf_vertical_4_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +#define vpx_lpf_vertical_4 vpx_lpf_vertical_4_neon + +void vpx_lpf_vertical_4_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1); +void vpx_lpf_vertical_4_dual_neon(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1); +#define vpx_lpf_vertical_4_dual vpx_lpf_vertical_4_dual_neon + +void vpx_lpf_vertical_8_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +void vpx_lpf_vertical_8_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh); +#define vpx_lpf_vertical_8 vpx_lpf_vertical_8_neon + +void vpx_lpf_vertical_8_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1); +void vpx_lpf_vertical_8_dual_neon(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1); +#define vpx_lpf_vertical_8_dual vpx_lpf_vertical_8_dual_neon + +void vpx_minmax_8x8_c(const uint8_t *s, int p, const uint8_t *d, int dp, int *min, int *max); +void vpx_minmax_8x8_neon(const uint8_t *s, int p, const uint8_t *d, int dp, int *min, int *max); +#define vpx_minmax_8x8 vpx_minmax_8x8_neon + +unsigned int vpx_mse16x16_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_mse16x16_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_mse16x16_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_mse16x16)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_mse16x8_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_mse16x8_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_mse16x8_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_mse16x8)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_mse8x16_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_mse8x16_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_mse8x16_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_mse8x16)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_mse8x8_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_mse8x8_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_mse8x8_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_mse8x8)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +void vpx_quantize_b_c(const tran_low_t *coeff_ptr, intptr_t n_coeffs, const struct macroblock_plane *const mb_plane, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const struct ScanOrder *const scan_order); +void vpx_quantize_b_neon(const tran_low_t *coeff_ptr, intptr_t n_coeffs, const struct macroblock_plane *const mb_plane, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const struct ScanOrder *const scan_order); +#define vpx_quantize_b vpx_quantize_b_neon + +void vpx_quantize_b_32x32_c(const tran_low_t *coeff_ptr, const struct macroblock_plane *const mb_plane, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const struct ScanOrder *const scan_order); +void vpx_quantize_b_32x32_neon(const tran_low_t *coeff_ptr, const struct macroblock_plane *const mb_plane, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const struct ScanOrder *const scan_order); +#define vpx_quantize_b_32x32 vpx_quantize_b_32x32_neon + +unsigned int vpx_sad16x16_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad16x16_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad16x16_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad16x16)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +unsigned int vpx_sad16x16_avg_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad16x16_avg_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad16x16_avg_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +RTCD_EXTERN unsigned int (*vpx_sad16x16_avg)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); + +void vpx_sad16x16x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad16x16x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad16x16x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad16x16x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad16x32_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad16x32_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad16x32_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad16x32)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +unsigned int vpx_sad16x32_avg_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad16x32_avg_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad16x32_avg_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +RTCD_EXTERN unsigned int (*vpx_sad16x32_avg)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); + +void vpx_sad16x32x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad16x32x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad16x32x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad16x32x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad16x8_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad16x8_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad16x8_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad16x8)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +unsigned int vpx_sad16x8_avg_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad16x8_avg_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad16x8_avg_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +RTCD_EXTERN unsigned int (*vpx_sad16x8_avg)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); + +void vpx_sad16x8x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad16x8x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad16x8x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad16x8x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad32x16_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad32x16_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad32x16_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad32x16)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +unsigned int vpx_sad32x16_avg_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad32x16_avg_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad32x16_avg_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +RTCD_EXTERN unsigned int (*vpx_sad32x16_avg)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); + +void vpx_sad32x16x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad32x16x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad32x16x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad32x16x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad32x32_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad32x32_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad32x32_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad32x32)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +unsigned int vpx_sad32x32_avg_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad32x32_avg_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad32x32_avg_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +RTCD_EXTERN unsigned int (*vpx_sad32x32_avg)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); + +void vpx_sad32x32x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad32x32x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad32x32x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad32x32x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad32x64_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad32x64_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad32x64_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad32x64)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +unsigned int vpx_sad32x64_avg_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad32x64_avg_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad32x64_avg_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +RTCD_EXTERN unsigned int (*vpx_sad32x64_avg)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); + +void vpx_sad32x64x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad32x64x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad32x64x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad32x64x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad4x4_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad4x4_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +#define vpx_sad4x4 vpx_sad4x4_neon + +unsigned int vpx_sad4x4_avg_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad4x4_avg_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +#define vpx_sad4x4_avg vpx_sad4x4_avg_neon + +void vpx_sad4x4x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad4x4x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +#define vpx_sad4x4x4d vpx_sad4x4x4d_neon + +unsigned int vpx_sad4x8_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad4x8_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +#define vpx_sad4x8 vpx_sad4x8_neon + +unsigned int vpx_sad4x8_avg_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad4x8_avg_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +#define vpx_sad4x8_avg vpx_sad4x8_avg_neon + +void vpx_sad4x8x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad4x8x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +#define vpx_sad4x8x4d vpx_sad4x8x4d_neon + +unsigned int vpx_sad64x32_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad64x32_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad64x32_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad64x32)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +unsigned int vpx_sad64x32_avg_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad64x32_avg_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad64x32_avg_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +RTCD_EXTERN unsigned int (*vpx_sad64x32_avg)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); + +void vpx_sad64x32x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad64x32x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad64x32x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad64x32x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad64x64_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad64x64_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad64x64_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad64x64)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +unsigned int vpx_sad64x64_avg_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad64x64_avg_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad64x64_avg_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +RTCD_EXTERN unsigned int (*vpx_sad64x64_avg)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); + +void vpx_sad64x64x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad64x64x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad64x64x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad64x64x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad8x16_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad8x16_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +#define vpx_sad8x16 vpx_sad8x16_neon + +unsigned int vpx_sad8x16_avg_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad8x16_avg_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +#define vpx_sad8x16_avg vpx_sad8x16_avg_neon + +void vpx_sad8x16x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad8x16x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +#define vpx_sad8x16x4d vpx_sad8x16x4d_neon + +unsigned int vpx_sad8x4_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad8x4_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +#define vpx_sad8x4 vpx_sad8x4_neon + +unsigned int vpx_sad8x4_avg_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad8x4_avg_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +#define vpx_sad8x4_avg vpx_sad8x4_avg_neon + +void vpx_sad8x4x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad8x4x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +#define vpx_sad8x4x4d vpx_sad8x4x4d_neon + +unsigned int vpx_sad8x8_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad8x8_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +#define vpx_sad8x8 vpx_sad8x8_neon + +unsigned int vpx_sad8x8_avg_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +unsigned int vpx_sad8x8_avg_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred); +#define vpx_sad8x8_avg vpx_sad8x8_avg_neon + +void vpx_sad8x8x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad8x8x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +#define vpx_sad8x8x4d vpx_sad8x8x4d_neon + +unsigned int vpx_sad_skip_16x16_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_16x16_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_16x16_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad_skip_16x16)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +void vpx_sad_skip_16x16x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_16x16x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_16x16x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad_skip_16x16x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad_skip_16x32_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_16x32_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_16x32_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad_skip_16x32)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +void vpx_sad_skip_16x32x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_16x32x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_16x32x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad_skip_16x32x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad_skip_16x8_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_16x8_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_16x8_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad_skip_16x8)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +void vpx_sad_skip_16x8x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_16x8x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_16x8x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad_skip_16x8x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad_skip_32x16_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_32x16_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_32x16_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad_skip_32x16)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +void vpx_sad_skip_32x16x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_32x16x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_32x16x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad_skip_32x16x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad_skip_32x32_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_32x32_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_32x32_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad_skip_32x32)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +void vpx_sad_skip_32x32x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_32x32x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_32x32x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad_skip_32x32x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad_skip_32x64_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_32x64_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_32x64_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad_skip_32x64)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +void vpx_sad_skip_32x64x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_32x64x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_32x64x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad_skip_32x64x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad_skip_4x4_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_4x4_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +#define vpx_sad_skip_4x4 vpx_sad_skip_4x4_neon + +void vpx_sad_skip_4x4x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_4x4x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +#define vpx_sad_skip_4x4x4d vpx_sad_skip_4x4x4d_neon + +unsigned int vpx_sad_skip_4x8_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_4x8_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +#define vpx_sad_skip_4x8 vpx_sad_skip_4x8_neon + +void vpx_sad_skip_4x8x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_4x8x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +#define vpx_sad_skip_4x8x4d vpx_sad_skip_4x8x4d_neon + +unsigned int vpx_sad_skip_64x32_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_64x32_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_64x32_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad_skip_64x32)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +void vpx_sad_skip_64x32x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_64x32x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_64x32x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad_skip_64x32x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad_skip_64x64_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_64x64_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_64x64_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +RTCD_EXTERN unsigned int (*vpx_sad_skip_64x64)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); + +void vpx_sad_skip_64x64x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_64x64x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_64x64x4d_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +RTCD_EXTERN void (*vpx_sad_skip_64x64x4d)(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); + +unsigned int vpx_sad_skip_8x16_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_8x16_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +#define vpx_sad_skip_8x16 vpx_sad_skip_8x16_neon + +void vpx_sad_skip_8x16x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_8x16x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +#define vpx_sad_skip_8x16x4d vpx_sad_skip_8x16x4d_neon + +unsigned int vpx_sad_skip_8x4_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_8x4_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +#define vpx_sad_skip_8x4 vpx_sad_skip_8x4_neon + +void vpx_sad_skip_8x4x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_8x4x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +#define vpx_sad_skip_8x4x4d vpx_sad_skip_8x4x4d_neon + +unsigned int vpx_sad_skip_8x8_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +unsigned int vpx_sad_skip_8x8_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride); +#define vpx_sad_skip_8x8 vpx_sad_skip_8x8_neon + +void vpx_sad_skip_8x8x4d_c(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +void vpx_sad_skip_8x8x4d_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *const ref_array[4], int ref_stride, uint32_t sad_array[4]); +#define vpx_sad_skip_8x8x4d vpx_sad_skip_8x8x4d_neon + +int vpx_satd_c(const int16_t *coeff, int length); +int vpx_satd_neon(const int16_t *coeff, int length); +#define vpx_satd vpx_satd_neon + +void vpx_scaled_2d_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +void vpx_scaled_2d_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +#define vpx_scaled_2d vpx_scaled_2d_neon + +void vpx_scaled_avg_2d_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +#define vpx_scaled_avg_2d vpx_scaled_avg_2d_c + +void vpx_scaled_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +#define vpx_scaled_avg_horiz vpx_scaled_avg_horiz_c + +void vpx_scaled_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +#define vpx_scaled_avg_vert vpx_scaled_avg_vert_c + +void vpx_scaled_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +#define vpx_scaled_horiz vpx_scaled_horiz_c + +void vpx_scaled_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h); +#define vpx_scaled_vert vpx_scaled_vert_c + +int64_t vpx_sse_c(const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, int width, int height); +int64_t vpx_sse_neon(const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, int width, int height); +int64_t vpx_sse_neon_dotprod(const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, int width, int height); +RTCD_EXTERN int64_t (*vpx_sse)(const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, int width, int height); + +uint32_t vpx_sub_pixel_avg_variance16x16_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +uint32_t vpx_sub_pixel_avg_variance16x16_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +#define vpx_sub_pixel_avg_variance16x16 vpx_sub_pixel_avg_variance16x16_neon + +uint32_t vpx_sub_pixel_avg_variance16x32_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +uint32_t vpx_sub_pixel_avg_variance16x32_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +#define vpx_sub_pixel_avg_variance16x32 vpx_sub_pixel_avg_variance16x32_neon + +uint32_t vpx_sub_pixel_avg_variance16x8_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +uint32_t vpx_sub_pixel_avg_variance16x8_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +#define vpx_sub_pixel_avg_variance16x8 vpx_sub_pixel_avg_variance16x8_neon + +uint32_t vpx_sub_pixel_avg_variance32x16_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +uint32_t vpx_sub_pixel_avg_variance32x16_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +#define vpx_sub_pixel_avg_variance32x16 vpx_sub_pixel_avg_variance32x16_neon + +uint32_t vpx_sub_pixel_avg_variance32x32_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +uint32_t vpx_sub_pixel_avg_variance32x32_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +#define vpx_sub_pixel_avg_variance32x32 vpx_sub_pixel_avg_variance32x32_neon + +uint32_t vpx_sub_pixel_avg_variance32x64_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +uint32_t vpx_sub_pixel_avg_variance32x64_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +#define vpx_sub_pixel_avg_variance32x64 vpx_sub_pixel_avg_variance32x64_neon + +uint32_t vpx_sub_pixel_avg_variance4x4_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +uint32_t vpx_sub_pixel_avg_variance4x4_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +#define vpx_sub_pixel_avg_variance4x4 vpx_sub_pixel_avg_variance4x4_neon + +uint32_t vpx_sub_pixel_avg_variance4x8_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +uint32_t vpx_sub_pixel_avg_variance4x8_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +#define vpx_sub_pixel_avg_variance4x8 vpx_sub_pixel_avg_variance4x8_neon + +uint32_t vpx_sub_pixel_avg_variance64x32_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +uint32_t vpx_sub_pixel_avg_variance64x32_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +#define vpx_sub_pixel_avg_variance64x32 vpx_sub_pixel_avg_variance64x32_neon + +uint32_t vpx_sub_pixel_avg_variance64x64_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +uint32_t vpx_sub_pixel_avg_variance64x64_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +#define vpx_sub_pixel_avg_variance64x64 vpx_sub_pixel_avg_variance64x64_neon + +uint32_t vpx_sub_pixel_avg_variance8x16_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +uint32_t vpx_sub_pixel_avg_variance8x16_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +#define vpx_sub_pixel_avg_variance8x16 vpx_sub_pixel_avg_variance8x16_neon + +uint32_t vpx_sub_pixel_avg_variance8x4_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +uint32_t vpx_sub_pixel_avg_variance8x4_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +#define vpx_sub_pixel_avg_variance8x4 vpx_sub_pixel_avg_variance8x4_neon + +uint32_t vpx_sub_pixel_avg_variance8x8_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +uint32_t vpx_sub_pixel_avg_variance8x8_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred); +#define vpx_sub_pixel_avg_variance8x8 vpx_sub_pixel_avg_variance8x8_neon + +uint32_t vpx_sub_pixel_variance16x16_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +uint32_t vpx_sub_pixel_variance16x16_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +#define vpx_sub_pixel_variance16x16 vpx_sub_pixel_variance16x16_neon + +uint32_t vpx_sub_pixel_variance16x32_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +uint32_t vpx_sub_pixel_variance16x32_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +#define vpx_sub_pixel_variance16x32 vpx_sub_pixel_variance16x32_neon + +uint32_t vpx_sub_pixel_variance16x8_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +uint32_t vpx_sub_pixel_variance16x8_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +#define vpx_sub_pixel_variance16x8 vpx_sub_pixel_variance16x8_neon + +uint32_t vpx_sub_pixel_variance32x16_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +uint32_t vpx_sub_pixel_variance32x16_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +#define vpx_sub_pixel_variance32x16 vpx_sub_pixel_variance32x16_neon + +uint32_t vpx_sub_pixel_variance32x32_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +uint32_t vpx_sub_pixel_variance32x32_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +#define vpx_sub_pixel_variance32x32 vpx_sub_pixel_variance32x32_neon + +uint32_t vpx_sub_pixel_variance32x64_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +uint32_t vpx_sub_pixel_variance32x64_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +#define vpx_sub_pixel_variance32x64 vpx_sub_pixel_variance32x64_neon + +uint32_t vpx_sub_pixel_variance4x4_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +uint32_t vpx_sub_pixel_variance4x4_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +#define vpx_sub_pixel_variance4x4 vpx_sub_pixel_variance4x4_neon + +uint32_t vpx_sub_pixel_variance4x8_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +uint32_t vpx_sub_pixel_variance4x8_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +#define vpx_sub_pixel_variance4x8 vpx_sub_pixel_variance4x8_neon + +uint32_t vpx_sub_pixel_variance64x32_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +uint32_t vpx_sub_pixel_variance64x32_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +#define vpx_sub_pixel_variance64x32 vpx_sub_pixel_variance64x32_neon + +uint32_t vpx_sub_pixel_variance64x64_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +uint32_t vpx_sub_pixel_variance64x64_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +#define vpx_sub_pixel_variance64x64 vpx_sub_pixel_variance64x64_neon + +uint32_t vpx_sub_pixel_variance8x16_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +uint32_t vpx_sub_pixel_variance8x16_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +#define vpx_sub_pixel_variance8x16 vpx_sub_pixel_variance8x16_neon + +uint32_t vpx_sub_pixel_variance8x4_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +uint32_t vpx_sub_pixel_variance8x4_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +#define vpx_sub_pixel_variance8x4 vpx_sub_pixel_variance8x4_neon + +uint32_t vpx_sub_pixel_variance8x8_c(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +uint32_t vpx_sub_pixel_variance8x8_neon(const uint8_t *src_ptr, int src_stride, int x_offset, int y_offset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse); +#define vpx_sub_pixel_variance8x8 vpx_sub_pixel_variance8x8_neon + +void vpx_subtract_block_c(int rows, int cols, int16_t *diff_ptr, ptrdiff_t diff_stride, const uint8_t *src_ptr, ptrdiff_t src_stride, const uint8_t *pred_ptr, ptrdiff_t pred_stride); +void vpx_subtract_block_neon(int rows, int cols, int16_t *diff_ptr, ptrdiff_t diff_stride, const uint8_t *src_ptr, ptrdiff_t src_stride, const uint8_t *pred_ptr, ptrdiff_t pred_stride); +#define vpx_subtract_block vpx_subtract_block_neon + +uint64_t vpx_sum_squares_2d_i16_c(const int16_t *src, int stride, int size); +uint64_t vpx_sum_squares_2d_i16_neon(const int16_t *src, int stride, int size); +uint64_t vpx_sum_squares_2d_i16_sve(const int16_t *src, int stride, int size); +RTCD_EXTERN uint64_t (*vpx_sum_squares_2d_i16)(const int16_t *src, int stride, int size); + +void vpx_tm_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_tm_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_tm_predictor_16x16 vpx_tm_predictor_16x16_neon + +void vpx_tm_predictor_32x32_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_tm_predictor_32x32_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_tm_predictor_32x32 vpx_tm_predictor_32x32_neon + +void vpx_tm_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_tm_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_tm_predictor_4x4 vpx_tm_predictor_4x4_neon + +void vpx_tm_predictor_8x8_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_tm_predictor_8x8_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_tm_predictor_8x8 vpx_tm_predictor_8x8_neon + +void vpx_v_predictor_16x16_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_v_predictor_16x16_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_v_predictor_16x16 vpx_v_predictor_16x16_neon + +void vpx_v_predictor_32x32_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_v_predictor_32x32_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_v_predictor_32x32 vpx_v_predictor_32x32_neon + +void vpx_v_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_v_predictor_4x4_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_v_predictor_4x4 vpx_v_predictor_4x4_neon + +void vpx_v_predictor_8x8_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +void vpx_v_predictor_8x8_neon(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_v_predictor_8x8 vpx_v_predictor_8x8_neon + +unsigned int vpx_variance16x16_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance16x16_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance16x16_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_variance16x16)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_variance16x32_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance16x32_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance16x32_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_variance16x32)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_variance16x8_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance16x8_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance16x8_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_variance16x8)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_variance32x16_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance32x16_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance32x16_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_variance32x16)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_variance32x32_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance32x32_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance32x32_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_variance32x32)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_variance32x64_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance32x64_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance32x64_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_variance32x64)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_variance4x4_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance4x4_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance4x4_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_variance4x4)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_variance4x8_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance4x8_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance4x8_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_variance4x8)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_variance64x32_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance64x32_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance64x32_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_variance64x32)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_variance64x64_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance64x64_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance64x64_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_variance64x64)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_variance8x16_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance8x16_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance8x16_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_variance8x16)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_variance8x4_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance8x4_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance8x4_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_variance8x4)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +unsigned int vpx_variance8x8_c(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance8x8_neon(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +unsigned int vpx_variance8x8_neon_dotprod(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); +RTCD_EXTERN unsigned int (*vpx_variance8x8)(const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse); + +void vpx_ve_predictor_4x4_c(uint8_t *dst, ptrdiff_t stride, const uint8_t *above, const uint8_t *left); +#define vpx_ve_predictor_4x4 vpx_ve_predictor_4x4_c + +int vpx_vector_var_c(const int16_t *ref, const int16_t *src, const int bwl); +int vpx_vector_var_neon(const int16_t *ref, const int16_t *src, const int bwl); +#define vpx_vector_var vpx_vector_var_neon + +void vpx_dsp_rtcd(void); + +#include "vpx_config.h" + +#ifdef RTCD_C +#include "vpx_ports/arm.h" +static void setup_rtcd_internal(void) +{ + int flags = arm_cpu_caps(); + + (void)flags; + + vpx_convolve8 = vpx_convolve8_neon; + if (flags & HAS_NEON_DOTPROD) vpx_convolve8 = vpx_convolve8_neon_dotprod; + if (flags & HAS_NEON_I8MM) vpx_convolve8 = vpx_convolve8_neon_i8mm; + vpx_convolve8_avg = vpx_convolve8_avg_neon; + if (flags & HAS_NEON_DOTPROD) vpx_convolve8_avg = vpx_convolve8_avg_neon_dotprod; + if (flags & HAS_NEON_I8MM) vpx_convolve8_avg = vpx_convolve8_avg_neon_i8mm; + vpx_convolve8_avg_horiz = vpx_convolve8_avg_horiz_neon; + if (flags & HAS_NEON_DOTPROD) vpx_convolve8_avg_horiz = vpx_convolve8_avg_horiz_neon_dotprod; + if (flags & HAS_NEON_I8MM) vpx_convolve8_avg_horiz = vpx_convolve8_avg_horiz_neon_i8mm; + vpx_convolve8_avg_vert = vpx_convolve8_avg_vert_neon; + if (flags & HAS_NEON_DOTPROD) vpx_convolve8_avg_vert = vpx_convolve8_avg_vert_neon_dotprod; + if (flags & HAS_NEON_I8MM) vpx_convolve8_avg_vert = vpx_convolve8_avg_vert_neon_i8mm; + vpx_convolve8_horiz = vpx_convolve8_horiz_neon; + if (flags & HAS_NEON_DOTPROD) vpx_convolve8_horiz = vpx_convolve8_horiz_neon_dotprod; + if (flags & HAS_NEON_I8MM) vpx_convolve8_horiz = vpx_convolve8_horiz_neon_i8mm; + vpx_convolve8_vert = vpx_convolve8_vert_neon; + if (flags & HAS_NEON_DOTPROD) vpx_convolve8_vert = vpx_convolve8_vert_neon_dotprod; + if (flags & HAS_NEON_I8MM) vpx_convolve8_vert = vpx_convolve8_vert_neon_i8mm; + vpx_get16x16var = vpx_get16x16var_neon; + if (flags & HAS_NEON_DOTPROD) vpx_get16x16var = vpx_get16x16var_neon_dotprod; + vpx_get4x4sse_cs = vpx_get4x4sse_cs_neon; + if (flags & HAS_NEON_DOTPROD) vpx_get4x4sse_cs = vpx_get4x4sse_cs_neon_dotprod; + vpx_get8x8var = vpx_get8x8var_neon; + if (flags & HAS_NEON_DOTPROD) vpx_get8x8var = vpx_get8x8var_neon_dotprod; + vpx_mse16x16 = vpx_mse16x16_neon; + if (flags & HAS_NEON_DOTPROD) vpx_mse16x16 = vpx_mse16x16_neon_dotprod; + vpx_mse16x8 = vpx_mse16x8_neon; + if (flags & HAS_NEON_DOTPROD) vpx_mse16x8 = vpx_mse16x8_neon_dotprod; + vpx_mse8x16 = vpx_mse8x16_neon; + if (flags & HAS_NEON_DOTPROD) vpx_mse8x16 = vpx_mse8x16_neon_dotprod; + vpx_mse8x8 = vpx_mse8x8_neon; + if (flags & HAS_NEON_DOTPROD) vpx_mse8x8 = vpx_mse8x8_neon_dotprod; + vpx_sad16x16 = vpx_sad16x16_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad16x16 = vpx_sad16x16_neon_dotprod; + vpx_sad16x16_avg = vpx_sad16x16_avg_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad16x16_avg = vpx_sad16x16_avg_neon_dotprod; + vpx_sad16x16x4d = vpx_sad16x16x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad16x16x4d = vpx_sad16x16x4d_neon_dotprod; + vpx_sad16x32 = vpx_sad16x32_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad16x32 = vpx_sad16x32_neon_dotprod; + vpx_sad16x32_avg = vpx_sad16x32_avg_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad16x32_avg = vpx_sad16x32_avg_neon_dotprod; + vpx_sad16x32x4d = vpx_sad16x32x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad16x32x4d = vpx_sad16x32x4d_neon_dotprod; + vpx_sad16x8 = vpx_sad16x8_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad16x8 = vpx_sad16x8_neon_dotprod; + vpx_sad16x8_avg = vpx_sad16x8_avg_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad16x8_avg = vpx_sad16x8_avg_neon_dotprod; + vpx_sad16x8x4d = vpx_sad16x8x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad16x8x4d = vpx_sad16x8x4d_neon_dotprod; + vpx_sad32x16 = vpx_sad32x16_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad32x16 = vpx_sad32x16_neon_dotprod; + vpx_sad32x16_avg = vpx_sad32x16_avg_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad32x16_avg = vpx_sad32x16_avg_neon_dotprod; + vpx_sad32x16x4d = vpx_sad32x16x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad32x16x4d = vpx_sad32x16x4d_neon_dotprod; + vpx_sad32x32 = vpx_sad32x32_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad32x32 = vpx_sad32x32_neon_dotprod; + vpx_sad32x32_avg = vpx_sad32x32_avg_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad32x32_avg = vpx_sad32x32_avg_neon_dotprod; + vpx_sad32x32x4d = vpx_sad32x32x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad32x32x4d = vpx_sad32x32x4d_neon_dotprod; + vpx_sad32x64 = vpx_sad32x64_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad32x64 = vpx_sad32x64_neon_dotprod; + vpx_sad32x64_avg = vpx_sad32x64_avg_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad32x64_avg = vpx_sad32x64_avg_neon_dotprod; + vpx_sad32x64x4d = vpx_sad32x64x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad32x64x4d = vpx_sad32x64x4d_neon_dotprod; + vpx_sad64x32 = vpx_sad64x32_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad64x32 = vpx_sad64x32_neon_dotprod; + vpx_sad64x32_avg = vpx_sad64x32_avg_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad64x32_avg = vpx_sad64x32_avg_neon_dotprod; + vpx_sad64x32x4d = vpx_sad64x32x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad64x32x4d = vpx_sad64x32x4d_neon_dotprod; + vpx_sad64x64 = vpx_sad64x64_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad64x64 = vpx_sad64x64_neon_dotprod; + vpx_sad64x64_avg = vpx_sad64x64_avg_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad64x64_avg = vpx_sad64x64_avg_neon_dotprod; + vpx_sad64x64x4d = vpx_sad64x64x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad64x64x4d = vpx_sad64x64x4d_neon_dotprod; + vpx_sad_skip_16x16 = vpx_sad_skip_16x16_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_16x16 = vpx_sad_skip_16x16_neon_dotprod; + vpx_sad_skip_16x16x4d = vpx_sad_skip_16x16x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_16x16x4d = vpx_sad_skip_16x16x4d_neon_dotprod; + vpx_sad_skip_16x32 = vpx_sad_skip_16x32_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_16x32 = vpx_sad_skip_16x32_neon_dotprod; + vpx_sad_skip_16x32x4d = vpx_sad_skip_16x32x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_16x32x4d = vpx_sad_skip_16x32x4d_neon_dotprod; + vpx_sad_skip_16x8 = vpx_sad_skip_16x8_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_16x8 = vpx_sad_skip_16x8_neon_dotprod; + vpx_sad_skip_16x8x4d = vpx_sad_skip_16x8x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_16x8x4d = vpx_sad_skip_16x8x4d_neon_dotprod; + vpx_sad_skip_32x16 = vpx_sad_skip_32x16_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_32x16 = vpx_sad_skip_32x16_neon_dotprod; + vpx_sad_skip_32x16x4d = vpx_sad_skip_32x16x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_32x16x4d = vpx_sad_skip_32x16x4d_neon_dotprod; + vpx_sad_skip_32x32 = vpx_sad_skip_32x32_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_32x32 = vpx_sad_skip_32x32_neon_dotprod; + vpx_sad_skip_32x32x4d = vpx_sad_skip_32x32x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_32x32x4d = vpx_sad_skip_32x32x4d_neon_dotprod; + vpx_sad_skip_32x64 = vpx_sad_skip_32x64_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_32x64 = vpx_sad_skip_32x64_neon_dotprod; + vpx_sad_skip_32x64x4d = vpx_sad_skip_32x64x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_32x64x4d = vpx_sad_skip_32x64x4d_neon_dotprod; + vpx_sad_skip_64x32 = vpx_sad_skip_64x32_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_64x32 = vpx_sad_skip_64x32_neon_dotprod; + vpx_sad_skip_64x32x4d = vpx_sad_skip_64x32x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_64x32x4d = vpx_sad_skip_64x32x4d_neon_dotprod; + vpx_sad_skip_64x64 = vpx_sad_skip_64x64_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_64x64 = vpx_sad_skip_64x64_neon_dotprod; + vpx_sad_skip_64x64x4d = vpx_sad_skip_64x64x4d_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sad_skip_64x64x4d = vpx_sad_skip_64x64x4d_neon_dotprod; + vpx_sse = vpx_sse_neon; + if (flags & HAS_NEON_DOTPROD) vpx_sse = vpx_sse_neon_dotprod; + vpx_sum_squares_2d_i16 = vpx_sum_squares_2d_i16_neon; + if (flags & HAS_SVE) vpx_sum_squares_2d_i16 = vpx_sum_squares_2d_i16_sve; + vpx_variance16x16 = vpx_variance16x16_neon; + if (flags & HAS_NEON_DOTPROD) vpx_variance16x16 = vpx_variance16x16_neon_dotprod; + vpx_variance16x32 = vpx_variance16x32_neon; + if (flags & HAS_NEON_DOTPROD) vpx_variance16x32 = vpx_variance16x32_neon_dotprod; + vpx_variance16x8 = vpx_variance16x8_neon; + if (flags & HAS_NEON_DOTPROD) vpx_variance16x8 = vpx_variance16x8_neon_dotprod; + vpx_variance32x16 = vpx_variance32x16_neon; + if (flags & HAS_NEON_DOTPROD) vpx_variance32x16 = vpx_variance32x16_neon_dotprod; + vpx_variance32x32 = vpx_variance32x32_neon; + if (flags & HAS_NEON_DOTPROD) vpx_variance32x32 = vpx_variance32x32_neon_dotprod; + vpx_variance32x64 = vpx_variance32x64_neon; + if (flags & HAS_NEON_DOTPROD) vpx_variance32x64 = vpx_variance32x64_neon_dotprod; + vpx_variance4x4 = vpx_variance4x4_neon; + if (flags & HAS_NEON_DOTPROD) vpx_variance4x4 = vpx_variance4x4_neon_dotprod; + vpx_variance4x8 = vpx_variance4x8_neon; + if (flags & HAS_NEON_DOTPROD) vpx_variance4x8 = vpx_variance4x8_neon_dotprod; + vpx_variance64x32 = vpx_variance64x32_neon; + if (flags & HAS_NEON_DOTPROD) vpx_variance64x32 = vpx_variance64x32_neon_dotprod; + vpx_variance64x64 = vpx_variance64x64_neon; + if (flags & HAS_NEON_DOTPROD) vpx_variance64x64 = vpx_variance64x64_neon_dotprod; + vpx_variance8x16 = vpx_variance8x16_neon; + if (flags & HAS_NEON_DOTPROD) vpx_variance8x16 = vpx_variance8x16_neon_dotprod; + vpx_variance8x4 = vpx_variance8x4_neon; + if (flags & HAS_NEON_DOTPROD) vpx_variance8x4 = vpx_variance8x4_neon_dotprod; + vpx_variance8x8 = vpx_variance8x8_neon; + if (flags & HAS_NEON_DOTPROD) vpx_variance8x8 = vpx_variance8x8_neon_dotprod; +} +#endif + +#ifdef __cplusplus +} // extern "C" +#endif + +#endif diff --git a/media/libvpx/config/mac/arm64/vpx_scale_rtcd.h b/media/libvpx/config/mac/arm64/vpx_scale_rtcd.h new file mode 100644 index 0000000000..b371368275 --- /dev/null +++ b/media/libvpx/config/mac/arm64/vpx_scale_rtcd.h @@ -0,0 +1,75 @@ +// This file is generated. Do not edit. +#ifndef VPX_SCALE_RTCD_H_ +#define VPX_SCALE_RTCD_H_ + +#ifdef RTCD_C +#define RTCD_EXTERN +#else +#define RTCD_EXTERN extern +#endif + +struct yv12_buffer_config; + +#ifdef __cplusplus +extern "C" { +#endif + +void vp8_horizontal_line_2_1_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width); +#define vp8_horizontal_line_2_1_scale vp8_horizontal_line_2_1_scale_c + +void vp8_horizontal_line_5_3_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width); +#define vp8_horizontal_line_5_3_scale vp8_horizontal_line_5_3_scale_c + +void vp8_horizontal_line_5_4_scale_c(const unsigned char *source, unsigned int source_width, unsigned char *dest, unsigned int dest_width); +#define vp8_horizontal_line_5_4_scale vp8_horizontal_line_5_4_scale_c + +void vp8_vertical_band_2_1_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width); +#define vp8_vertical_band_2_1_scale vp8_vertical_band_2_1_scale_c + +void vp8_vertical_band_2_1_scale_i_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width); +#define vp8_vertical_band_2_1_scale_i vp8_vertical_band_2_1_scale_i_c + +void vp8_vertical_band_5_3_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width); +#define vp8_vertical_band_5_3_scale vp8_vertical_band_5_3_scale_c + +void vp8_vertical_band_5_4_scale_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width); +#define vp8_vertical_band_5_4_scale vp8_vertical_band_5_4_scale_c + +void vp8_yv12_copy_frame_c(const struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc); +#define vp8_yv12_copy_frame vp8_yv12_copy_frame_c + +void vp8_yv12_extend_frame_borders_c(struct yv12_buffer_config *ybf); +#define vp8_yv12_extend_frame_borders vp8_yv12_extend_frame_borders_c + +void vpx_extend_frame_borders_c(struct yv12_buffer_config *ybf); +#define vpx_extend_frame_borders vpx_extend_frame_borders_c + +void vpx_extend_frame_inner_borders_c(struct yv12_buffer_config *ybf); +#define vpx_extend_frame_inner_borders vpx_extend_frame_inner_borders_c + +void vpx_yv12_copy_frame_c(const struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc); +#define vpx_yv12_copy_frame vpx_yv12_copy_frame_c + +void vpx_yv12_copy_y_c(const struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc); +#define vpx_yv12_copy_y vpx_yv12_copy_y_c + +void vpx_scale_rtcd(void); + +#include "vpx_config.h" + +#ifdef RTCD_C +#include "vpx_ports/arm.h" +static void setup_rtcd_internal(void) +{ + int flags = arm_cpu_caps(); + + (void)flags; + +} +#endif + +#ifdef __cplusplus +} // extern "C" +#endif + +#endif diff --git a/media/libvpx/generate_sources_mozbuild.sh b/media/libvpx/generate_sources_mozbuild.sh index 4efcb54aa1..f84dc75be1 100755 --- a/media/libvpx/generate_sources_mozbuild.sh +++ b/media/libvpx/generate_sources_mozbuild.sh @@ -69,6 +69,7 @@ function write_sources { # Convert a list of source files into sources.mozbuild. # $1 - Input file. # $2 - Output prefix. +# $3 - Path of vpx_config.c under $LIBVPX_CONFIG_DIR function convert_srcs_to_project_files { # Do the following here: # 1. Filter .c, .h, .s, .S and .asm files. @@ -76,9 +77,17 @@ function convert_srcs_to_project_files { local source_list=$(grep -E '(\.c|\.h|\.S|\.s|\.asm)$' $1) - # Remove vpx_config.c. - # The platform-specific vpx_config.c will be added into in moz.build later. + # Adjust the path for vpx_config.c while maintaining list order: + # Since the config file resides in $BASE_DIR/$LIBVPX_CONFIG_DIR, while the + # files in $source_list are placed under $BASE_DIR/libvpx (see write_sources), + # the config file path requires adjustment. To ensure the list remains sorted, + # we must first remove it and then insert it at the beginning of the list. + + # Remove vpx_config.c source_list=$(echo "$source_list" | grep -v 'vpx_config\.c') + # Insert vpx_config.c at the beginning of the list. + local config=$(echo "../$LIBVPX_CONFIG_DIR/$3/vpx_config.c") + source_list=$(echo "$config" ; echo "$source_list") # Remove include-only asm files (no object code emitted) source_list=$(echo "$source_list" | grep -v 'x86_abi_support\.asm') @@ -169,6 +178,7 @@ function gen_rtcd_header { # $1 - Header file directory. # $2 - Config command line. function gen_config_files { + mkdir -p $BASE_DIR/$LIBVPX_CONFIG_DIR/$1 ./configure $2 --log=$BASE_DIR/$LIBVPX_CONFIG_DIR/$1/config.log > /dev/null echo "Log file: $BASE_DIR/$LIBVPX_CONFIG_DIR/$1/config.log" @@ -202,9 +212,9 @@ all_platforms="--enable-external-build --disable-examples --disable-install-docs all_platforms="${all_platforms} --enable-multi-res-encoding --size-limit=8192x4608 --enable-pic" all_platforms="${all_platforms} --disable-avx512" x86_platforms="--enable-postproc --enable-vp9-postproc --as=yasm" -arm_platforms="--enable-runtime-cpu-detect --enable-realtime-only" -arm64_platforms="--enable-realtime-only" -disable_sve="--disable-sve" # Bug 1885585 +runtime_cpu_detect="--enable-runtime-cpu-detect" +realtime_only="--enable-realtime-only" +disable_sve="--disable-sve" # Bug 1885585, Bug 1889813 gen_config_files linux/x64 "--target=x86_64-linux-gcc ${all_platforms} ${x86_platforms}" gen_config_files linux/ia32 "--target=x86-linux-gcc ${all_platforms} ${x86_platforms}" @@ -213,9 +223,10 @@ gen_config_files mac/ia32 "--target=x86-darwin9-gcc ${all_platforms} ${x86_platf gen_config_files win/x64 "--target=x86_64-win64-vs15 ${all_platforms} ${x86_platforms}" gen_config_files win/ia32 "--target=x86-win32-gcc ${all_platforms} ${x86_platforms}" -gen_config_files linux/arm "--target=armv7-linux-gcc ${all_platforms} ${arm_platforms}" -gen_config_files linux/arm64 "--target=arm64-linux-gcc ${all_platforms} ${arm64_platforms}" -gen_config_files win/aarch64 "--target=arm64-win64-vs15 ${all_platforms} ${arm64_platforms} ${disable_sve}" # Bug 1885585 +gen_config_files linux/arm "--target=armv7-linux-gcc ${all_platforms} ${runtime_cpu_detect} ${realtime_only}" +gen_config_files linux/arm64 "--target=arm64-linux-gcc ${all_platforms} ${realtime_only} ${disable_sve}" # Bug 1889813 +gen_config_files mac/arm64 "--target=arm64-darwin-gcc ${all_platforms}" +gen_config_files win/aarch64 "--target=arm64-win64-vs15 ${all_platforms} ${realtime_only} ${disable_sve}" # Bug 1885585 gen_config_files generic "--target=generic-gnu ${all_platforms}" @@ -237,7 +248,8 @@ gen_rtcd_header win/x64 x86_64 gen_rtcd_header win/ia32 x86 gen_rtcd_header linux/arm armv7 -gen_rtcd_header linux/arm64 arm64 +gen_rtcd_header linux/arm64 arm64 $disable_sve # Bug 1889813 +gen_rtcd_header mac/arm64 arm64 gen_rtcd_header win/aarch64 arm64 $disable_sve # Bug 1885585 gen_rtcd_header generic generic @@ -251,39 +263,74 @@ rm -rf $BASE_DIR/sources.mozbuild write_license $BASE_DIR/sources.mozbuild echo "files = {" >> $BASE_DIR/sources.mozbuild -echo "Generate X86_64 source list." +echo "Generate X86_64 source list on Linux." config=$(print_config linux/x64) make_clean make libvpx_srcs.txt target=libs $config > /dev/null -convert_srcs_to_project_files libvpx_srcs.txt X64 +convert_srcs_to_project_files libvpx_srcs.txt LINUX_X64 linux/x64 + +echo "Generate X86_64 source list on Mac." +config=$(print_config mac/x64) +make_clean +make libvpx_srcs.txt target=libs $config > /dev/null +convert_srcs_to_project_files libvpx_srcs.txt MAC_X64 mac/x64 + +echo "Generate X86_64 source list on Windows." +config=$(print_config win/x64) +make_clean +make libvpx_srcs.txt target=libs $config > /dev/null +convert_srcs_to_project_files libvpx_srcs.txt WIN_X64 win/x64 # Copy vpx_version.h once. The file is the same for all platforms. cp vpx_version.h $BASE_DIR/$LIBVPX_CONFIG_DIR -echo "Generate IA32 source list." +echo "Generate IA32 source list on Linux." config=$(print_config linux/ia32) make_clean make libvpx_srcs.txt target=libs $config > /dev/null -convert_srcs_to_project_files libvpx_srcs.txt IA32 +convert_srcs_to_project_files libvpx_srcs.txt LINUX_IA32 linux/ia32 + +echo "Generate IA32 source list on Mac." +config=$(print_config mac/ia32) +make_clean +make libvpx_srcs.txt target=libs $config > /dev/null +convert_srcs_to_project_files libvpx_srcs.txt MAC_IA32 mac/ia32 -echo "Generate ARM source list." +echo "Generate IA32 source list on Windows." +config=$(print_config win/ia32) +make_clean +make libvpx_srcs.txt target=libs $config > /dev/null +convert_srcs_to_project_files libvpx_srcs.txt WIN_IA32 win/ia32 + +echo "Generate ARM source list on Linux." config=$(print_config linux/arm) make_clean make libvpx_srcs.txt target=libs $config > /dev/null -convert_srcs_to_project_files libvpx_srcs.txt ARM +convert_srcs_to_project_files libvpx_srcs.txt LINUX_ARM linux/arm -echo "Generate ARM64 source list." +echo "Generate ARM64 source list on Linux" config=$(print_config linux/arm64) make_clean make libvpx_srcs.txt target=libs $config > /dev/null -convert_srcs_to_project_files libvpx_srcs.txt ARM64 -# Bug 1885585: The sve files will be excluded from the win/aarch64 build in moz.build. +convert_srcs_to_project_files libvpx_srcs.txt LINUX_ARM64 linux/arm64 + +echo "Generate ARM64 source list on Mac" +config=$(print_config mac/arm64) +make_clean +make libvpx_srcs.txt target=libs $config > /dev/null +convert_srcs_to_project_files libvpx_srcs.txt MAC_ARM64 mac/arm64 + +echo "Generate AARCH64 source list on Windows." +config=$(print_config win/aarch64) +make_clean +make libvpx_srcs.txt target=libs $config > /dev/null +convert_srcs_to_project_files libvpx_srcs.txt WIN_AARCH64 win/aarch64 echo "Generate generic source list." config=$(print_config generic) make_clean make libvpx_srcs.txt target=libs $config > /dev/null -convert_srcs_to_project_files libvpx_srcs.txt GENERIC +convert_srcs_to_project_files libvpx_srcs.txt GENERIC generic echo "}" >> $BASE_DIR/sources.mozbuild diff --git a/media/libvpx/libvpx/build/make/rtcd.pl b/media/libvpx/libvpx/build/make/rtcd.pl index 025238d678..d84966d83b 100755 --- a/media/libvpx/libvpx/build/make/rtcd.pl +++ b/media/libvpx/libvpx/build/make/rtcd.pl @@ -73,6 +73,10 @@ sub vpx_config($) { } sub specialize { + if (@_ <= 1) { + die "'specialize' must be called with a function name and at least one ", + "architecture ('C' is implied): \n@_\n"; + } my $fn=$_[0]; shift; foreach my $opt (@_) { diff --git a/media/libvpx/libvpx/test/add_noise_test.cc b/media/libvpx/libvpx/test/add_noise_test.cc index 4fc4e81e63..8f1c3be041 100644 --- a/media/libvpx/libvpx/test/add_noise_test.cc +++ b/media/libvpx/libvpx/test/add_noise_test.cc @@ -16,6 +16,7 @@ #include "third_party/googletest/src/include/gtest/gtest.h" #include "./vpx_dsp_rtcd.h" #include "vpx/vpx_integer.h" +#include "vpx_config.h" #include "vpx_dsp/postproc.h" #include "vpx_mem/vpx_mem.h" diff --git a/media/libvpx/libvpx/test/altref_test.cc b/media/libvpx/libvpx/test/altref_test.cc index 903230fde9..edd2e58361 100644 --- a/media/libvpx/libvpx/test/altref_test.cc +++ b/media/libvpx/libvpx/test/altref_test.cc @@ -12,6 +12,7 @@ #include "test/encode_test_driver.h" #include "test/i420_video_source.h" #include "test/util.h" +#include "vpx_config.h" namespace { #if CONFIG_VP8_ENCODER diff --git a/media/libvpx/libvpx/test/bench.cc b/media/libvpx/libvpx/test/bench.cc index 4b883d8250..0783f2a734 100644 --- a/media/libvpx/libvpx/test/bench.cc +++ b/media/libvpx/libvpx/test/bench.cc @@ -10,6 +10,7 @@ #include <stdio.h> #include <algorithm> +#include <cstdlib> #include "test/bench.h" #include "vpx_ports/vpx_timer.h" diff --git a/media/libvpx/libvpx/test/borders_test.cc b/media/libvpx/libvpx/test/borders_test.cc index 2726bd557d..cf8fbd9c83 100644 --- a/media/libvpx/libvpx/test/borders_test.cc +++ b/media/libvpx/libvpx/test/borders_test.cc @@ -14,6 +14,7 @@ #include "test/encode_test_driver.h" #include "test/i420_video_source.h" #include "test/util.h" +#include "vpx_config.h" namespace { @@ -79,6 +80,11 @@ TEST_P(BordersTest, TestLowBitrate) { ASSERT_NO_FATAL_FAILURE(RunLoop(&video)); } +#if CONFIG_REALTIME_ONLY +VP9_INSTANTIATE_TEST_SUITE(BordersTest, + ::testing::Values(::libvpx_test::kRealTime)); +#else VP9_INSTANTIATE_TEST_SUITE(BordersTest, ::testing::Values(::libvpx_test::kTwoPassGood)); +#endif } // namespace diff --git a/media/libvpx/libvpx/test/comp_avg_pred_test.cc b/media/libvpx/libvpx/test/comp_avg_pred_test.cc index 3234cc9a25..2e8fb5c61d 100644 --- a/media/libvpx/libvpx/test/comp_avg_pred_test.cc +++ b/media/libvpx/libvpx/test/comp_avg_pred_test.cc @@ -15,6 +15,7 @@ #include "test/acm_random.h" #include "test/buffer.h" #include "test/register_state_check.h" +#include "vpx_config.h" #include "vpx_ports/vpx_timer.h" namespace { diff --git a/media/libvpx/libvpx/test/convolve_test.cc b/media/libvpx/libvpx/test/convolve_test.cc index 11f7625137..7594bb77e5 100644 --- a/media/libvpx/libvpx/test/convolve_test.cc +++ b/media/libvpx/libvpx/test/convolve_test.cc @@ -1228,10 +1228,16 @@ WRAP(convolve8_avg_horiz_sve, 12) #endif // HAVE_SVE #if HAVE_SVE2 +WRAP(convolve8_sve2, 8) +WRAP(convolve8_avg_sve2, 8) WRAP(convolve8_vert_sve2, 8) WRAP(convolve8_avg_vert_sve2, 8) +WRAP(convolve8_sve2, 10) +WRAP(convolve8_avg_sve2, 10) WRAP(convolve8_vert_sve2, 10) WRAP(convolve8_avg_vert_sve2, 10) +WRAP(convolve8_sve2, 12) +WRAP(convolve8_avg_sve2, 12) WRAP(convolve8_vert_sve2, 12) WRAP(convolve8_avg_vert_sve2, 12) #endif // HAVE_SVE2 @@ -1495,23 +1501,23 @@ INSTANTIATE_TEST_SUITE_P(SVE, ConvolveTest, const ConvolveFunctions convolve8_sve2( wrap_convolve_copy_c_8, wrap_convolve_avg_c_8, wrap_convolve8_horiz_c_8, wrap_convolve8_avg_horiz_c_8, wrap_convolve8_vert_sve2_8, - wrap_convolve8_avg_vert_sve2_8, wrap_convolve8_c_8, wrap_convolve8_avg_c_8, - wrap_convolve8_horiz_c_8, wrap_convolve8_avg_horiz_c_8, - wrap_convolve8_vert_c_8, wrap_convolve8_avg_vert_c_8, wrap_convolve8_c_8, - wrap_convolve8_avg_c_8, 8); + wrap_convolve8_avg_vert_sve2_8, wrap_convolve8_sve2_8, + wrap_convolve8_avg_sve2_8, wrap_convolve8_horiz_c_8, + wrap_convolve8_avg_horiz_c_8, wrap_convolve8_vert_c_8, + wrap_convolve8_avg_vert_c_8, wrap_convolve8_c_8, wrap_convolve8_avg_c_8, 8); const ConvolveFunctions convolve10_sve2( wrap_convolve_copy_c_10, wrap_convolve_avg_c_10, wrap_convolve8_horiz_c_10, wrap_convolve8_avg_horiz_c_10, wrap_convolve8_vert_sve2_10, - wrap_convolve8_avg_vert_sve2_10, wrap_convolve8_c_10, - wrap_convolve8_avg_c_10, wrap_convolve8_horiz_c_10, + wrap_convolve8_avg_vert_sve2_10, wrap_convolve8_sve2_10, + wrap_convolve8_avg_sve2_10, wrap_convolve8_horiz_c_10, wrap_convolve8_avg_horiz_c_10, wrap_convolve8_vert_c_10, wrap_convolve8_avg_vert_c_10, wrap_convolve8_c_10, wrap_convolve8_avg_c_10, 10); const ConvolveFunctions convolve12_sve2( wrap_convolve_copy_c_12, wrap_convolve_avg_c_12, wrap_convolve8_horiz_c_12, wrap_convolve8_avg_horiz_c_12, wrap_convolve8_vert_sve2_12, - wrap_convolve8_avg_vert_sve2_12, wrap_convolve8_c_12, - wrap_convolve8_avg_c_12, wrap_convolve8_horiz_c_12, + wrap_convolve8_avg_vert_sve2_12, wrap_convolve8_sve2_12, + wrap_convolve8_avg_sve2_12, wrap_convolve8_horiz_c_12, wrap_convolve8_avg_horiz_c_12, wrap_convolve8_vert_c_12, wrap_convolve8_avg_vert_c_12, wrap_convolve8_c_12, wrap_convolve8_avg_c_12, 12); diff --git a/media/libvpx/libvpx/test/cpu_speed_test.cc b/media/libvpx/libvpx/test/cpu_speed_test.cc index 22f4552963..ee00cc4cf2 100644 --- a/media/libvpx/libvpx/test/cpu_speed_test.cc +++ b/media/libvpx/libvpx/test/cpu_speed_test.cc @@ -148,9 +148,6 @@ TEST_P(CpuSpeedTest, TestLowBitrate) { ASSERT_NO_FATAL_FAILURE(RunLoop(&video)); } -VP9_INSTANTIATE_TEST_SUITE(CpuSpeedTest, - ::testing::Values(::libvpx_test::kTwoPassGood, - ::libvpx_test::kOnePassGood, - ::libvpx_test::kRealTime), +VP9_INSTANTIATE_TEST_SUITE(CpuSpeedTest, ONE_PASS_TEST_MODES, ::testing::Range(0, 10)); } // namespace diff --git a/media/libvpx/libvpx/test/cq_test.cc b/media/libvpx/libvpx/test/cq_test.cc index b74915a336..4ccdf8538b 100644 --- a/media/libvpx/libvpx/test/cq_test.cc +++ b/media/libvpx/libvpx/test/cq_test.cc @@ -14,6 +14,7 @@ #include "test/encode_test_driver.h" #include "test/i420_video_source.h" #include "test/util.h" +#include "vpx_config.h" namespace { @@ -104,6 +105,10 @@ CQTest::BitrateMap CQTest::bitrates_; TEST_P(CQTest, LinearPSNRIsHigherForCQLevel) { const vpx_rational timebase = { 33333333, 1000000000 }; +#if CONFIG_REALTIME_ONlY + GTEST_SKIP() + << "Non-zero g_lag_in_frames is unsupported with CONFIG_REALTIME_ONLY"; +#else cfg_.g_timebase = timebase; cfg_.rc_target_bitrate = kCQTargetBitrate; cfg_.g_lag_in_frames = 25; @@ -124,6 +129,7 @@ TEST_P(CQTest, LinearPSNRIsHigherForCQLevel) { ASSERT_NO_FATAL_FAILURE(RunLoop(&video)); const double vbr_psnr_lin = GetLinearPSNROverBitrate(); EXPECT_GE(cq_psnr_lin, vbr_psnr_lin); +#endif // CONFIG_REALTIME_ONLY } VP8_INSTANTIATE_TEST_SUITE(CQTest, ::testing::Range(kCQLevelMin, kCQLevelMax, diff --git a/media/libvpx/libvpx/test/dct16x16_test.cc b/media/libvpx/libvpx/test/dct16x16_test.cc index 8c4213ee16..f5e1cbb15b 100644 --- a/media/libvpx/libvpx/test/dct16x16_test.cc +++ b/media/libvpx/libvpx/test/dct16x16_test.cc @@ -25,6 +25,7 @@ #include "vp9/common/vp9_scan.h" #include "vpx/vpx_codec.h" #include "vpx/vpx_integer.h" +#include "vpx_config.h" #include "vpx_ports/mem.h" #include "vpx_ports/msvc.h" // for round() #include "vpx_ports/vpx_timer.h" diff --git a/media/libvpx/libvpx/test/dct_partial_test.cc b/media/libvpx/libvpx/test/dct_partial_test.cc index ec6f543f71..d9844c6c48 100644 --- a/media/libvpx/libvpx/test/dct_partial_test.cc +++ b/media/libvpx/libvpx/test/dct_partial_test.cc @@ -22,6 +22,7 @@ #include "test/clear_system_state.h" #include "test/register_state_check.h" #include "test/util.h" +#include "vpx_config.h" #include "vpx/vpx_codec.h" #include "vpx/vpx_integer.h" #include "vpx_dsp/vpx_dsp_common.h" diff --git a/media/libvpx/libvpx/test/dct_test.cc b/media/libvpx/libvpx/test/dct_test.cc index c3d3081c42..d5ae81b498 100644 --- a/media/libvpx/libvpx/test/dct_test.cc +++ b/media/libvpx/libvpx/test/dct_test.cc @@ -23,6 +23,7 @@ #include "test/register_state_check.h" #include "test/util.h" #include "vp9/common/vp9_entropy.h" +#include "vpx_config.h" #include "vpx/vpx_codec.h" #include "vpx/vpx_integer.h" #include "vpx_ports/mem.h" diff --git a/media/libvpx/libvpx/test/decode_corrupted.cc b/media/libvpx/libvpx/test/decode_corrupted.cc index 58773d7b86..a03bf3ba86 100644 --- a/media/libvpx/libvpx/test/decode_corrupted.cc +++ b/media/libvpx/libvpx/test/decode_corrupted.cc @@ -16,6 +16,7 @@ #include "test/encode_test_driver.h" #include "test/util.h" #include "test/i420_video_source.h" +#include "vpx_config.h" #include "vpx_mem/vpx_mem.h" namespace { diff --git a/media/libvpx/libvpx/test/encode_api_test.cc b/media/libvpx/libvpx/test/encode_api_test.cc index ca3b17a5d5..155b83544a 100644 --- a/media/libvpx/libvpx/test/encode_api_test.cc +++ b/media/libvpx/libvpx/test/encode_api_test.cc @@ -847,6 +847,10 @@ TEST(EncodeAPI, PtsOrDurationTooBig) { // 32-bit systems defined by VPX_MAX_ALLOCABLE_MEMORY #if VPX_ARCH_X86_64 || VPX_ARCH_AARCH64 TEST(EncodeAPI, ConfigLargeTargetBitrateVp9) { +#ifdef CHROMIUM + GTEST_SKIP() << "Under Chromium's configuration the allocator is unable" + "to provide the space required for the frames below."; +#else constexpr int kWidth = 12383; constexpr int kHeight = 8192; constexpr auto *iface = &vpx_codec_vp9_cx_algo; @@ -868,6 +872,7 @@ TEST(EncodeAPI, ConfigLargeTargetBitrateVp9) { EXPECT_NO_FATAL_FAILURE(InitCodec(*iface, kWidth, kHeight, &enc.ctx, &cfg)) << "target bitrate: " << cfg.rc_target_bitrate << " framerate: " << static_cast<double>(cfg.g_timebase.den) / cfg.g_timebase.num; +#endif // defined(CHROMIUM) } #endif // VPX_ARCH_X86_64 || VPX_ARCH_AARCH64 @@ -875,12 +880,14 @@ TEST(EncodeAPI, ConfigLargeTargetBitrateVp9) { class VP9Encoder { public: explicit VP9Encoder(int speed) - : speed_(speed), bit_depth_(VPX_BITS_8), fmt_(VPX_IMG_FMT_I420) {} + : speed_(speed), row_mt_(0), bit_depth_(VPX_BITS_8), + fmt_(VPX_IMG_FMT_I420) {} // The image format `fmt` must not have the VPX_IMG_FMT_HIGHBITDEPTH bit set. // If bit_depth > 8, we will set the VPX_IMG_FMT_HIGHBITDEPTH bit before // passing the image format to vpx_img_alloc(). - VP9Encoder(int speed, vpx_bit_depth_t bit_depth, vpx_img_fmt_t fmt) - : speed_(speed), bit_depth_(bit_depth), fmt_(fmt) {} + VP9Encoder(int speed, unsigned int row_mt, vpx_bit_depth_t bit_depth, + vpx_img_fmt_t fmt) + : speed_(speed), row_mt_(row_mt), bit_depth_(bit_depth), fmt_(fmt) {} ~VP9Encoder(); void Configure(unsigned int threads, unsigned int width, unsigned int height, @@ -889,6 +896,7 @@ class VP9Encoder { private: const int speed_; + const unsigned int row_mt_; const vpx_bit_depth_t bit_depth_; const vpx_img_fmt_t fmt_; bool initialized_ = false; @@ -938,6 +946,7 @@ void VP9Encoder::Configure(unsigned int threads, unsigned int width, high_bit_depth ? VPX_CODEC_USE_HIGHBITDEPTH : 0), VPX_CODEC_OK); ASSERT_EQ(vpx_codec_control(&enc_, VP8E_SET_CPUUSED, speed_), VPX_CODEC_OK); + ASSERT_EQ(vpx_codec_control(&enc_, VP9E_SET_ROW_MT, row_mt_), VPX_CODEC_OK); initialized_ = true; return; } @@ -1312,6 +1321,145 @@ TEST(EncodeAPI, Buganizer319964497) { encoder.Encode(/*key_frame=*/false); } +TEST(EncodeAPI, Buganizer329088759RowMT0) { + VP9Encoder encoder(8, 0, VPX_BITS_8, VPX_IMG_FMT_I444); + encoder.Configure(/*threads=*/8, /*width=*/1686, /*height=*/398, VPX_VBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/true); + encoder.Encode(/*key_frame=*/false); + encoder.Configure(/*threads=*/0, /*width=*/1686, /*height=*/1, VPX_VBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/true); + encoder.Configure(/*threads=*/0, /*width=*/1482, /*height=*/113, VPX_CBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/true); + encoder.Configure(/*threads=*/0, /*width=*/881, /*height=*/59, VPX_CBR, + VPX_DL_REALTIME); + encoder.Configure(/*threads=*/13, /*width=*/1271, /*height=*/385, VPX_CBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/false); + encoder.Configure(/*threads=*/2, /*width=*/1, /*height=*/62, VPX_VBR, + VPX_DL_REALTIME); +} + +TEST(EncodeAPI, Buganizer329088759RowMT1) { + VP9Encoder encoder(8, 1, VPX_BITS_8, VPX_IMG_FMT_I444); + encoder.Configure(/*threads=*/8, /*width=*/1686, /*height=*/398, VPX_VBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/true); + encoder.Encode(/*key_frame=*/false); + // Needs to set threads to non-zero to repro the issue. + encoder.Configure(/*threads=*/2, /*width=*/1686, /*height=*/1, VPX_VBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/true); + encoder.Configure(/*threads=*/2, /*width=*/1482, /*height=*/113, VPX_CBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/true); + encoder.Configure(/*threads=*/2, /*width=*/881, /*height=*/59, VPX_CBR, + VPX_DL_REALTIME); + encoder.Configure(/*threads=*/13, /*width=*/1271, /*height=*/385, VPX_CBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/false); + encoder.Configure(/*threads=*/2, /*width=*/1, /*height=*/62, VPX_VBR, + VPX_DL_REALTIME); +} + +TEST(EncodeAPI, Buganizer331086799) { + VP9Encoder encoder(6, 1, VPX_BITS_8, VPX_IMG_FMT_I420); + encoder.Configure(0, 1385, 1, VPX_CBR, VPX_DL_REALTIME); + encoder.Configure(0, 1, 1, VPX_VBR, VPX_DL_REALTIME); + encoder.Encode(false); + encoder.Configure(16, 1385, 1, VPX_VBR, VPX_DL_GOOD_QUALITY); + encoder.Encode(false); + encoder.Encode(false); + encoder.Configure(0, 1, 1, VPX_CBR, VPX_DL_REALTIME); + encoder.Encode(true); +} + +TEST(EncodeAPI, Buganizer331108729) { + VP9Encoder encoder(1, 1, VPX_BITS_8, VPX_IMG_FMT_I422); + encoder.Configure(0, 1919, 260, VPX_VBR, VPX_DL_REALTIME); + encoder.Configure(9, 440, 1, VPX_CBR, VPX_DL_GOOD_QUALITY); + encoder.Encode(true); + encoder.Configure(8, 1919, 260, VPX_VBR, VPX_DL_REALTIME); + encoder.Encode(false); +} + +TEST(EncodeAPI, Buganizer331108922BitDepth8) { + VP9Encoder encoder(9, 1, VPX_BITS_8, VPX_IMG_FMT_I420); + encoder.Configure(/*threads=*/1, /*width=*/1, /*height=*/1080, VPX_VBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/false); + encoder.Configure(/*threads=*/0, /*width=*/1, /*height=*/1080, VPX_CBR, + VPX_DL_GOOD_QUALITY); + encoder.Configure(/*threads=*/16, /*width=*/1, /*height=*/394, VPX_CBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/false); + encoder.Encode(/*key_frame=*/true); + encoder.Configure(/*threads=*/16, /*width=*/1, /*height=*/798, VPX_CBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/false); +} + +#if CONFIG_VP9_HIGHBITDEPTH +TEST(EncodeAPI, Buganizer329674887RowMT0BitDepth12) { + VP9Encoder encoder(8, 0, VPX_BITS_12, VPX_IMG_FMT_I444); + encoder.Configure(/*threads=*/2, /*width=*/1030, /*height=*/583, VPX_VBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/true); + encoder.Configure(/*threads=*/0, /*width=*/1030, /*height=*/1, VPX_CBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/true); + encoder.Configure(/*threads=*/0, /*width=*/548, /*height=*/322, VPX_VBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/false); + encoder.Configure(/*threads=*/16, /*width=*/24, /*height=*/583, VPX_CBR, + VPX_DL_GOOD_QUALITY); +} + +TEST(EncodeAPI, Buganizer329179808RowMT0BitDepth10) { + VP9Encoder encoder(4, 0, VPX_BITS_10, VPX_IMG_FMT_I444); + encoder.Configure(/*threads=*/16, /*width=*/1488, /*height=*/5, VPX_VBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/true); + encoder.Configure(/*threads=*/16, /*width=*/839, /*height=*/1, VPX_CBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/false); + encoder.Configure(/*threads=*/11, /*width=*/657, /*height=*/5, VPX_CBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/false); +} + +TEST(EncodeAPI, Buganizer329179808RowMT1BitDepth10) { + VP9Encoder encoder(4, 1, VPX_BITS_10, VPX_IMG_FMT_I444); + encoder.Configure(/*threads=*/16, /*width=*/1488, /*height=*/5, VPX_VBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/true); + encoder.Configure(/*threads=*/16, /*width=*/839, /*height=*/1, VPX_CBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/false); + encoder.Configure(/*threads=*/11, /*width=*/657, /*height=*/5, VPX_CBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/false); +} + +TEST(EncodeAPI, Buganizer331108922BitDepth12) { + VP9Encoder encoder(9, 1, VPX_BITS_12, VPX_IMG_FMT_I444); + encoder.Configure(/*threads=*/1, /*width=*/1, /*height=*/1080, VPX_VBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/false); + encoder.Configure(/*threads=*/0, /*width=*/1, /*height=*/1080, VPX_CBR, + VPX_DL_GOOD_QUALITY); + encoder.Configure(/*threads=*/16, /*width=*/1, /*height=*/394, VPX_CBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/false); + encoder.Encode(/*key_frame=*/true); + encoder.Configure(/*threads=*/16, /*width=*/1, /*height=*/798, VPX_CBR, + VPX_DL_REALTIME); + encoder.Encode(/*key_frame=*/false); +} +#endif + #endif // CONFIG_VP9_ENCODER } // namespace diff --git a/media/libvpx/libvpx/test/encode_test_driver.h b/media/libvpx/libvpx/test/encode_test_driver.h index 7dd80d6664..76a5718021 100644 --- a/media/libvpx/libvpx/test/encode_test_driver.h +++ b/media/libvpx/libvpx/test/encode_test_driver.h @@ -33,15 +33,24 @@ enum TestMode { kTwoPassGood, kTwoPassBest }; + +#if CONFIG_REALTIME_ONLY +#define ALL_TEST_MODES ::testing::Values(::libvpx_test::kRealTime) +#define ONE_PASS_TEST_MODES ::testing::Values(::libvpx_test::kRealTime) +#define ONE_OR_TWO_PASS_TEST_MODES ::testing::Values(::libvpx_test::kRealTime) +#else #define ALL_TEST_MODES \ ::testing::Values(::libvpx_test::kRealTime, ::libvpx_test::kOnePassGood, \ ::libvpx_test::kOnePassBest, ::libvpx_test::kTwoPassGood, \ ::libvpx_test::kTwoPassBest) - #define ONE_PASS_TEST_MODES \ ::testing::Values(::libvpx_test::kRealTime, ::libvpx_test::kOnePassGood, \ ::libvpx_test::kOnePassBest) +#define ONE_OR_TWO_PASS_TEST_MODES \ + ::testing::Values(::libvpx_test::kOnePassGood, ::libvpx_test::kTwoPassGood) +#endif + #define TWO_PASS_TEST_MODES \ ::testing::Values(::libvpx_test::kTwoPassGood, ::libvpx_test::kTwoPassBest) diff --git a/media/libvpx/libvpx/test/error_resilience_test.cc b/media/libvpx/libvpx/test/error_resilience_test.cc index 6b019b2bfb..2fbf98fde0 100644 --- a/media/libvpx/libvpx/test/error_resilience_test.cc +++ b/media/libvpx/libvpx/test/error_resilience_test.cc @@ -13,6 +13,7 @@ #include "test/encode_test_driver.h" #include "test/i420_video_source.h" #include "test/util.h" +#include "vpx_config.h" namespace { @@ -194,6 +195,10 @@ class ErrorResilienceTestLarge }; TEST_P(ErrorResilienceTestLarge, OnVersusOff) { +#if CONFIG_REALTIME_ONLY + GTEST_SKIP() + << "Non-zero g_lag_in_frames is unsupported with CONFIG_REALTIME_ONLY"; +#else const vpx_rational timebase = { 33333333, 1000000000 }; cfg_.g_timebase = timebase; cfg_.rc_target_bitrate = 2000; @@ -222,6 +227,7 @@ TEST_P(ErrorResilienceTestLarge, OnVersusOff) { EXPECT_GE(psnr_ratio, 0.9); EXPECT_LE(psnr_ratio, 1.1); } +#endif // CONFIG_REALTIME_ONLY } // Check for successful decoding and no encoder/decoder mismatch diff --git a/media/libvpx/libvpx/test/fdct8x8_test.cc b/media/libvpx/libvpx/test/fdct8x8_test.cc index 3cdf909d46..042f96b0a6 100644 --- a/media/libvpx/libvpx/test/fdct8x8_test.cc +++ b/media/libvpx/libvpx/test/fdct8x8_test.cc @@ -23,6 +23,7 @@ #include "test/util.h" #include "vp9/common/vp9_entropy.h" #include "vp9/common/vp9_scan.h" +#include "vpx_config.h" #include "vpx/vpx_codec.h" #include "vpx/vpx_integer.h" #include "vpx_ports/mem.h" diff --git a/media/libvpx/libvpx/test/frame_size_tests.cc b/media/libvpx/libvpx/test/frame_size_tests.cc index 6306e4f2ca..235f34857a 100644 --- a/media/libvpx/libvpx/test/frame_size_tests.cc +++ b/media/libvpx/libvpx/test/frame_size_tests.cc @@ -13,6 +13,7 @@ #include "test/codec_factory.h" #include "test/register_state_check.h" #include "test/video_source.h" +#include "vpx_config.h" namespace { @@ -167,6 +168,9 @@ class VP9FrameSizeTestsLarge : public ::libvpx_test::EncoderTest, }; TEST_F(VP9FrameSizeTestsLarge, TestInvalidSizes) { +#ifdef CHROMIUM + GTEST_SKIP() << "16K framebuffers are not supported by Chromium's allocator."; +#else ::libvpx_test::RandomVideoSource video; #if CONFIG_SIZE_LIMIT @@ -175,9 +179,16 @@ TEST_F(VP9FrameSizeTestsLarge, TestInvalidSizes) { expected_res_ = VPX_CODEC_MEM_ERROR; ASSERT_NO_FATAL_FAILURE(RunLoop(&video, expected_res_)); #endif + +#endif } TEST_F(VP9FrameSizeTestsLarge, ValidSizes) { +#ifdef CHROMIUM + GTEST_SKIP() + << "Under Chromium's configuration the allocator is unable to provide" + "the space required for a single frame at the maximum resolution."; +#else ::libvpx_test::RandomVideoSource video; #if CONFIG_SIZE_LIMIT @@ -202,6 +213,8 @@ TEST_F(VP9FrameSizeTestsLarge, ValidSizes) { expected_res_ = VPX_CODEC_OK; ASSERT_NO_FATAL_FAILURE(::libvpx_test::EncoderTest::RunLoop(&video)); #endif + +#endif // defined(CHROMIUM) } TEST_F(VP9FrameSizeTestsLarge, OneByOneVideo) { diff --git a/media/libvpx/libvpx/test/hadamard_test.cc b/media/libvpx/libvpx/test/hadamard_test.cc index b22bae87cc..5e50984418 100644 --- a/media/libvpx/libvpx/test/hadamard_test.cc +++ b/media/libvpx/libvpx/test/hadamard_test.cc @@ -17,6 +17,7 @@ #include "test/acm_random.h" #include "test/register_state_check.h" +#include "vpx_config.h" namespace { diff --git a/media/libvpx/libvpx/test/level_test.cc b/media/libvpx/libvpx/test/level_test.cc index 36cfd645c9..1f2f904698 100644 --- a/media/libvpx/libvpx/test/level_test.cc +++ b/media/libvpx/libvpx/test/level_test.cc @@ -12,6 +12,7 @@ #include "test/encode_test_driver.h" #include "test/i420_video_source.h" #include "test/util.h" +#include "vpx_config.h" namespace { class LevelTest @@ -67,6 +68,9 @@ class LevelTest }; TEST_P(LevelTest, TestTargetLevel11Large) { +#if CONFIG_REALTIME_ONLY + GTEST_SKIP(); +#else ASSERT_NE(encoding_mode_, ::libvpx_test::kRealTime); ::libvpx_test::I420VideoSource video("hantro_odd.yuv", 208, 144, 30, 1, 0, 60); @@ -74,9 +78,13 @@ TEST_P(LevelTest, TestTargetLevel11Large) { cfg_.rc_target_bitrate = 150; ASSERT_NO_FATAL_FAILURE(RunLoop(&video)); ASSERT_GE(target_level_, level_); +#endif } TEST_P(LevelTest, TestTargetLevel20Large) { +#if CONFIG_REALTIME_ONLY + GTEST_SKIP(); +#else ASSERT_NE(encoding_mode_, ::libvpx_test::kRealTime); ::libvpx_test::I420VideoSource video("hantro_collage_w352h288.yuv", 352, 288, 30, 1, 0, 60); @@ -84,9 +92,13 @@ TEST_P(LevelTest, TestTargetLevel20Large) { cfg_.rc_target_bitrate = 1200; ASSERT_NO_FATAL_FAILURE(RunLoop(&video)); ASSERT_GE(target_level_, level_); +#endif } TEST_P(LevelTest, TestTargetLevel31Large) { +#if CONFIG_REALTIME_ONLY + GTEST_SKIP(); +#else ASSERT_NE(encoding_mode_, ::libvpx_test::kRealTime); ::libvpx_test::I420VideoSource video("niklas_1280_720_30.y4m", 1280, 720, 30, 1, 0, 60); @@ -94,6 +106,7 @@ TEST_P(LevelTest, TestTargetLevel31Large) { cfg_.rc_target_bitrate = 8000; ASSERT_NO_FATAL_FAILURE(RunLoop(&video)); ASSERT_GE(target_level_, level_); +#endif } // Test for keeping level stats only @@ -140,8 +153,6 @@ TEST_P(LevelTest, TestTargetLevelApi) { EXPECT_EQ(VPX_CODEC_OK, vpx_codec_destroy(&enc)); } -VP9_INSTANTIATE_TEST_SUITE(LevelTest, - ::testing::Values(::libvpx_test::kTwoPassGood, - ::libvpx_test::kOnePassGood), +VP9_INSTANTIATE_TEST_SUITE(LevelTest, ONE_OR_TWO_PASS_TEST_MODES, ::testing::Range(0, 9)); } // namespace diff --git a/media/libvpx/libvpx/test/minmax_test.cc b/media/libvpx/libvpx/test/minmax_test.cc index b495709063..2d616ad15f 100644 --- a/media/libvpx/libvpx/test/minmax_test.cc +++ b/media/libvpx/libvpx/test/minmax_test.cc @@ -13,6 +13,7 @@ #include "third_party/googletest/src/include/gtest/gtest.h" +#include "vpx_config.h" #include "./vpx_dsp_rtcd.h" #include "vpx/vpx_integer.h" #include "vpx_mem/vpx_mem.h" diff --git a/media/libvpx/libvpx/test/partial_idct_test.cc b/media/libvpx/libvpx/test/partial_idct_test.cc index 01e63eb691..4cf0b2b02e 100644 --- a/media/libvpx/libvpx/test/partial_idct_test.cc +++ b/media/libvpx/libvpx/test/partial_idct_test.cc @@ -25,6 +25,7 @@ #include "vp9/common/vp9_blockd.h" #include "vp9/common/vp9_scan.h" #include "vpx/vpx_integer.h" +#include "vpx_config.h" #include "vpx_ports/vpx_timer.h" using libvpx_test::ACMRandom; diff --git a/media/libvpx/libvpx/test/realtime_test.cc b/media/libvpx/libvpx/test/realtime_test.cc index a9870b3cbf..b5ab7f8ce2 100644 --- a/media/libvpx/libvpx/test/realtime_test.cc +++ b/media/libvpx/libvpx/test/realtime_test.cc @@ -14,6 +14,7 @@ #include "test/util.h" #include "test/video_source.h" #include "third_party/googletest/src/include/gtest/gtest.h" +#include "vpx_config.h" namespace { @@ -94,6 +95,9 @@ TEST_P(RealtimeTest, RealtimeDefaultCpuUsed) { TEST_P(RealtimeTest, IntegerOverflow) { TestIntegerOverflow(2048, 2048); } TEST_P(RealtimeTest, IntegerOverflowLarge) { +#ifdef CHROMIUM + GTEST_SKIP() << "16K framebuffers are not supported by Chromium's allocator."; +#else if (IsVP9()) { #if VPX_ARCH_AARCH64 || VPX_ARCH_X86_64 TestIntegerOverflow(16384, 16384); @@ -107,6 +111,7 @@ TEST_P(RealtimeTest, IntegerOverflowLarge) { "warnings are fixed."; // TestIntegerOverflow(16383, 16383); } +#endif // defined(CHROMIUM) } VP8_INSTANTIATE_TEST_SUITE(RealtimeTest, diff --git a/media/libvpx/libvpx/test/resize_test.cc b/media/libvpx/libvpx/test/resize_test.cc index f27bd7ebbc..cecd058f51 100644 --- a/media/libvpx/libvpx/test/resize_test.cc +++ b/media/libvpx/libvpx/test/resize_test.cc @@ -15,6 +15,7 @@ #include "test/i420_video_source.h" #include "test/video_source.h" #include "test/util.h" +#include "vpx_config.h" // Enable(1) or Disable(0) writing of the compressed bitstream. #define WRITE_COMPRESSED_STREAM 0 diff --git a/media/libvpx/libvpx/test/test.mk b/media/libvpx/libvpx/test/test.mk index d4521f08bf..28fe9dbb8d 100644 --- a/media/libvpx/libvpx/test/test.mk +++ b/media/libvpx/libvpx/test/test.mk @@ -21,9 +21,12 @@ LIBVPX_TEST_SRCS-yes += video_source.h ## Black box tests only use the public API. ## LIBVPX_TEST_SRCS-yes += ../md5_utils.h ../md5_utils.c +LIBVPX_TEST_SRCS-yes += vpx_image_test.cc LIBVPX_TEST_SRCS-$(CONFIG_DECODERS) += ivf_video_source.h LIBVPX_TEST_SRCS-$(CONFIG_ENCODERS) += ../y4minput.h ../y4minput.c +ifneq ($(CONFIG_REALTIME_ONLY),yes) LIBVPX_TEST_SRCS-$(CONFIG_ENCODERS) += altref_test.cc +endif LIBVPX_TEST_SRCS-$(CONFIG_ENCODERS) += encode_api_test.cc LIBVPX_TEST_SRCS-$(CONFIG_ENCODERS) += error_resilience_test.cc LIBVPX_TEST_SRCS-$(CONFIG_ENCODERS) += i420_video_source.h @@ -32,7 +35,9 @@ LIBVPX_TEST_SRCS-$(CONFIG_ENCODERS) += resize_test.cc LIBVPX_TEST_SRCS-$(CONFIG_ENCODERS) += y4m_video_source.h LIBVPX_TEST_SRCS-$(CONFIG_ENCODERS) += yuv_video_source.h +ifneq ($(CONFIG_REALTIME_ONLY),yes) LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += config_test.cc +endif LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += cq_test.cc LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += keyframe_test.cc LIBVPX_TEST_SRCS-$(CONFIG_VP8_ENCODER) += vp8_datarate_test.cc @@ -43,7 +48,9 @@ LIBVPX_TEST_SRCS-$(CONFIG_VP9_DECODER) += external_frame_buffer_test.cc LIBVPX_TEST_SRCS-$(CONFIG_VP9_DECODER) += user_priv_test.cc LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += active_map_refresh_test.cc LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += active_map_test.cc +ifneq ($(CONFIG_REALTIME_ONLY),yes) LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += alt_ref_aq_segment_test.cc +endif LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += aq_segment_test.cc LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += borders_test.cc LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += cpu_speed_test.cc @@ -60,7 +67,9 @@ LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += svc_test.h LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += svc_end_to_end_test.cc LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += timestamp_test.cc LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += vp9_datarate_test.cc +ifneq ($(CONFIG_REALTIME_ONLY),yes) LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += vp9_ext_ratectrl_test.cc +endif LIBVPX_TEST_SRCS-$(CONFIG_VP9_ENCODER) += ../vp9/simple_encode.h LIBVPX_TEST_SRCS-yes += decode_test_driver.cc diff --git a/media/libvpx/libvpx/test/test_vectors.cc b/media/libvpx/libvpx/test/test_vectors.cc index 3ffc3efc41..954ff771a9 100644 --- a/media/libvpx/libvpx/test/test_vectors.cc +++ b/media/libvpx/libvpx/test/test_vectors.cc @@ -9,6 +9,7 @@ */ #include "test/test_vectors.h" +#include "vpx_config.h" namespace libvpx_test { diff --git a/media/libvpx/libvpx/test/timestamp_test.cc b/media/libvpx/libvpx/test/timestamp_test.cc index 00abf8f31c..da9e81952c 100644 --- a/media/libvpx/libvpx/test/timestamp_test.cc +++ b/media/libvpx/libvpx/test/timestamp_test.cc @@ -12,6 +12,7 @@ #include "test/util.h" #include "test/video_source.h" #include "third_party/googletest/src/include/gtest/gtest.h" +#include "vpx_config.h" namespace { @@ -94,8 +95,15 @@ TEST_P(TimestampTest, TestVpxRollover) { ASSERT_NO_FATAL_FAILURE(RunLoop(&video)); } +#if CONFIG_REALTIME_ONLY +VP8_INSTANTIATE_TEST_SUITE(TimestampTest, + ::testing::Values(::libvpx_test::kRealTime)); +VP9_INSTANTIATE_TEST_SUITE(TimestampTest, + ::testing::Values(::libvpx_test::kRealTime)); +#else VP8_INSTANTIATE_TEST_SUITE(TimestampTest, ::testing::Values(::libvpx_test::kTwoPassGood)); VP9_INSTANTIATE_TEST_SUITE(TimestampTest, ::testing::Values(::libvpx_test::kTwoPassGood)); +#endif } // namespace diff --git a/media/libvpx/libvpx/test/vp8_denoiser_sse2_test.cc b/media/libvpx/libvpx/test/vp8_denoiser_sse2_test.cc index 7fa867d8bb..0918049f0f 100644 --- a/media/libvpx/libvpx/test/vp8_denoiser_sse2_test.cc +++ b/media/libvpx/libvpx/test/vp8_denoiser_sse2_test.cc @@ -21,6 +21,7 @@ #include "vp8/encoder/denoising.h" #include "vp8/common/reconinter.h" #include "vpx/vpx_integer.h" +#include "vpx_config.h" #include "vpx_mem/vpx_mem.h" using libvpx_test::ACMRandom; diff --git a/media/libvpx/libvpx/test/vp9_arf_freq_test.cc b/media/libvpx/libvpx/test/vp9_arf_freq_test.cc index 3882326d2f..14ce96ce25 100644 --- a/media/libvpx/libvpx/test/vp9_arf_freq_test.cc +++ b/media/libvpx/libvpx/test/vp9_arf_freq_test.cc @@ -18,6 +18,7 @@ #include "test/y4m_video_source.h" #include "test/yuv_video_source.h" #include "vp9/encoder/vp9_ratectrl.h" +#include "vpx_config.h" namespace { diff --git a/media/libvpx/libvpx/test/vp9_boolcoder_test.cc b/media/libvpx/libvpx/test/vp9_boolcoder_test.cc index 6ba171a000..aeff0d7a50 100644 --- a/media/libvpx/libvpx/test/vp9_boolcoder_test.cc +++ b/media/libvpx/libvpx/test/vp9_boolcoder_test.cc @@ -53,7 +53,7 @@ TEST(VP9, TestBitIO) { ACMRandom bit_rnd(random_seed); vpx_writer bw; uint8_t bw_buffer[kBufferSize]; - vpx_start_encode(&bw, bw_buffer); + vpx_start_encode(&bw, bw_buffer, sizeof(bw_buffer)); int bit = (bit_method == 0) ? 0 : (bit_method == 1) ? 1 : 0; for (int i = 0; i < kBitsToTest; ++i) { @@ -65,7 +65,7 @@ TEST(VP9, TestBitIO) { vpx_write(&bw, bit, static_cast<int>(probas[i])); } - vpx_stop_encode(&bw); + GTEST_ASSERT_EQ(vpx_stop_encode(&bw), 0); // vpx_reader_fill() may read into uninitialized data that // isn't used meaningfully, but may trigger an MSan warning. memset(bw_buffer + bw.pos, 0, sizeof(BD_VALUE) - 1); @@ -90,3 +90,24 @@ TEST(VP9, TestBitIO) { } } } + +TEST(VP9, TestBitIOBufferSize0) { + vpx_writer bw; + uint8_t bw_buffer[1]; + vpx_start_encode(&bw, bw_buffer, 0); + GTEST_ASSERT_EQ(vpx_stop_encode(&bw), -1); +} + +TEST(VP9, TestBitIOBufferSize1) { + vpx_writer bw; + uint8_t bw_buffer[1]; + vpx_start_encode(&bw, bw_buffer, sizeof(bw_buffer)); + GTEST_ASSERT_EQ(vpx_stop_encode(&bw), -1); +} + +TEST(VP9, TestBitIOBufferSize2) { + vpx_writer bw; + uint8_t bw_buffer[2]; + vpx_start_encode(&bw, bw_buffer, sizeof(bw_buffer)); + GTEST_ASSERT_EQ(vpx_stop_encode(&bw), 0); +} diff --git a/media/libvpx/libvpx/test/vp9_denoiser_test.cc b/media/libvpx/libvpx/test/vp9_denoiser_test.cc index 831f83305c..b5a2c1c0a2 100644 --- a/media/libvpx/libvpx/test/vp9_denoiser_test.cc +++ b/media/libvpx/libvpx/test/vp9_denoiser_test.cc @@ -24,6 +24,7 @@ #include "vp9/common/vp9_reconinter.h" #include "vp9/encoder/vp9_context_tree.h" #include "vp9/encoder/vp9_denoiser.h" +#include "vpx_config.h" using libvpx_test::ACMRandom; diff --git a/media/libvpx/libvpx/test/vp9_end_to_end_test.cc b/media/libvpx/libvpx/test/vp9_end_to_end_test.cc index 79be4ee146..0ed48ef051 100644 --- a/media/libvpx/libvpx/test/vp9_end_to_end_test.cc +++ b/media/libvpx/libvpx/test/vp9_end_to_end_test.cc @@ -18,6 +18,7 @@ #include "test/util.h" #include "test/y4m_video_source.h" #include "test/yuv_video_source.h" +#include "vpx_config.h" namespace { @@ -65,7 +66,9 @@ const TestVideoParam kTestVectorsNv12[] = { // Encoding modes tested const libvpx_test::TestMode kEncodingModeVectors[] = { +#if !CONFIG_REALTIME_ONLY ::libvpx_test::kTwoPassGood, ::libvpx_test::kOnePassGood, +#endif ::libvpx_test::kRealTime }; diff --git a/media/libvpx/libvpx/test/vp9_ethread_test.cc b/media/libvpx/libvpx/test/vp9_ethread_test.cc index c8d3cba7fb..c3f288aabb 100644 --- a/media/libvpx/libvpx/test/vp9_ethread_test.cc +++ b/media/libvpx/libvpx/test/vp9_ethread_test.cc @@ -17,6 +17,7 @@ #include "test/util.h" #include "test/y4m_video_source.h" #include "vp9/encoder/vp9_firstpass.h" +#include "vpx_config.h" namespace { // FIRSTPASS_STATS struct: @@ -168,6 +169,9 @@ static void compare_fp_stats_md5(vpx_fixed_buf_t *fp_stats) { } TEST_P(VPxFirstPassEncoderThreadTest, FirstPassStatsTest) { +#if CONFIG_REALTIME_ONLY + GTEST_SKIP(); +#else ::libvpx_test::Y4mVideoSource video("niklas_1280_720_30.y4m", 0, 60); first_pass_only_ = true; @@ -216,6 +220,7 @@ TEST_P(VPxFirstPassEncoderThreadTest, FirstPassStatsTest) { // Compare to check if stats match with row-mt=0/1. compare_fp_stats_md5(&firstpass_stats_); +#endif // CONFIG_REALTIME_ONLY } class VPxEncoderThreadTest @@ -407,23 +412,17 @@ INSTANTIATE_TEST_SUITE_P( ::testing::Combine( ::testing::Values( static_cast<const libvpx_test::CodecFactory *>(&libvpx_test::kVP9)), - ::testing::Values(::libvpx_test::kTwoPassGood, - ::libvpx_test::kOnePassGood, - ::libvpx_test::kRealTime), - ::testing::Range(3, 10), // cpu_used - ::testing::Range(0, 3), // tile_columns - ::testing::Range(2, 5))); // threads + ONE_PASS_TEST_MODES, ::testing::Range(3, 10), // cpu_used + ::testing::Range(0, 3), // tile_columns + ::testing::Range(2, 5))); // threads INSTANTIATE_TEST_SUITE_P( VP9Large, VPxEncoderThreadTest, ::testing::Combine( ::testing::Values( static_cast<const libvpx_test::CodecFactory *>(&libvpx_test::kVP9)), - ::testing::Values(::libvpx_test::kTwoPassGood, - ::libvpx_test::kOnePassGood, - ::libvpx_test::kRealTime), - ::testing::Range(0, 3), // cpu_used - ::testing::Range(0, 3), // tile_columns - ::testing::Range(2, 5))); // threads + ONE_PASS_TEST_MODES, ::testing::Range(0, 3), // cpu_used + ::testing::Range(0, 3), // tile_columns + ::testing::Range(2, 5))); // threads } // namespace diff --git a/media/libvpx/libvpx/test/vp9_ext_ratectrl_test.cc b/media/libvpx/libvpx/test/vp9_ext_ratectrl_test.cc index 5c23a5b0d5..3f3fc23854 100644 --- a/media/libvpx/libvpx/test/vp9_ext_ratectrl_test.cc +++ b/media/libvpx/libvpx/test/vp9_ext_ratectrl_test.cc @@ -51,6 +51,22 @@ class RateControllerForTest { gop_decision.use_key_frame = current_gop_ == 0 ? 1 : 0; gop_decision.use_alt_ref = 1; gop_decision.gop_coding_frames = kFixedGOPSize; + // First frame is key frame + gop_decision.update_type[0] = VPX_RC_KF_UPDATE; + for (int i = 1; i < kFixedGOPSize; i++) { + gop_decision.update_type[i] = VPX_RC_LF_UPDATE; + gop_decision.update_ref_index[i] = 0; + gop_decision.ref_frame_list[i].index[0] = 0; + gop_decision.ref_frame_list[i].name[0] = VPX_RC_LAST_FRAME; + gop_decision.ref_frame_list[i].index[1] = 0; + gop_decision.ref_frame_list[i].name[1] = VPX_RC_GOLDEN_FRAME; + gop_decision.ref_frame_list[i].index[2] = 0; + gop_decision.ref_frame_list[i].name[1] = VPX_RC_ALTREF_FRAME; + } + + // Second frame is altref + gop_decision.update_type[1] = VPX_RC_ARF_UPDATE; + gop_decision.update_ref_index[1] = 2; return gop_decision; } @@ -136,7 +152,11 @@ class ExtRateCtrlTest : public ::libvpx_test::EncoderTest, void SetUp() override { InitializeConfig(); +#if CONFIG_REALTIME_ONLY + SetMode(::libvpx_test::kRealTime); +#else SetMode(::libvpx_test::kTwoPassGood); +#endif } void PreEncodeFrameHook(::libvpx_test::VideoSource *video, diff --git a/media/libvpx/libvpx/test/vp9_lossless_test.cc b/media/libvpx/libvpx/test/vp9_lossless_test.cc index fe3cd1aba4..a955132f01 100644 --- a/media/libvpx/libvpx/test/vp9_lossless_test.cc +++ b/media/libvpx/libvpx/test/vp9_lossless_test.cc @@ -118,8 +118,13 @@ TEST_P(LosslessTest, TestLossLessEncodingCtrl) { EXPECT_GE(psnr_lossless, kMaxPsnr); } +#if CONFIG_REALTIME_ONLY +VP9_INSTANTIATE_TEST_SUITE(LosslessTest, + ::testing::Values(::libvpx_test::kRealTime)); +#else VP9_INSTANTIATE_TEST_SUITE(LosslessTest, ::testing::Values(::libvpx_test::kRealTime, ::libvpx_test::kOnePassGood, ::libvpx_test::kTwoPassGood)); +#endif } // namespace diff --git a/media/libvpx/libvpx/test/vp9_motion_vector_test.cc b/media/libvpx/libvpx/test/vp9_motion_vector_test.cc index 495ea11fce..37111e2461 100644 --- a/media/libvpx/libvpx/test/vp9_motion_vector_test.cc +++ b/media/libvpx/libvpx/test/vp9_motion_vector_test.cc @@ -16,6 +16,7 @@ #include "test/encode_test_driver.h" #include "test/util.h" #include "test/yuv_video_source.h" +#include "vpx_config.h" namespace { #define MAX_EXTREME_MV 1 @@ -23,7 +24,9 @@ namespace { // Encoding modes const libvpx_test::TestMode kEncodingModeVectors[] = { +#if !CONFIG_REALTIME_ONLY ::libvpx_test::kTwoPassGood, ::libvpx_test::kOnePassGood, +#endif ::libvpx_test::kRealTime }; diff --git a/media/libvpx/libvpx/test/vpx_image_test.cc b/media/libvpx/libvpx/test/vpx_image_test.cc new file mode 100644 index 0000000000..3d24b239a8 --- /dev/null +++ b/media/libvpx/libvpx/test/vpx_image_test.cc @@ -0,0 +1,127 @@ +/* + * Copyright (c) 2024 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#include <climits> + +#include "vpx/vpx_image.h" +#include "third_party/googletest/src/include/gtest/gtest.h" + +TEST(VpxImageTest, VpxImgWrapInvalidAlign) { + const int kWidth = 128; + const int kHeight = 128; + unsigned char buf[kWidth * kHeight * 3]; + + vpx_image_t img; + // Set img_data and img_data_owner to junk values. vpx_img_wrap() should + // not read these values on failure. + unsigned char empty[] = ""; + img.img_data = empty; + img.img_data_owner = 1; + + vpx_img_fmt_t format = VPX_IMG_FMT_I444; + // 'align' must be a power of 2 but is not. This causes the vpx_img_wrap() + // call to fail. The test verifies we do not read the junk values in 'img'. + unsigned int align = 31; + EXPECT_EQ(vpx_img_wrap(&img, format, kWidth, kHeight, align, buf), nullptr); +} + +TEST(VpxImageTest, VpxImgSetRectOverflow) { + const int kWidth = 128; + const int kHeight = 128; + unsigned char buf[kWidth * kHeight * 3]; + + vpx_image_t img; + vpx_img_fmt_t format = VPX_IMG_FMT_I444; + unsigned int align = 32; + EXPECT_EQ(vpx_img_wrap(&img, format, kWidth, kHeight, align, buf), &img); + + EXPECT_EQ(vpx_img_set_rect(&img, 0, 0, kWidth, kHeight), 0); + // This would result in overflow because -1 is cast to UINT_MAX. + EXPECT_NE(vpx_img_set_rect(&img, static_cast<unsigned int>(-1), + static_cast<unsigned int>(-1), kWidth, kHeight), + 0); +} + +TEST(VpxImageTest, VpxImgAllocNone) { + const int kWidth = 128; + const int kHeight = 128; + + vpx_image_t img; + vpx_img_fmt_t format = VPX_IMG_FMT_NONE; + unsigned int align = 32; + ASSERT_EQ(vpx_img_alloc(&img, format, kWidth, kHeight, align), nullptr); +} + +TEST(VpxImageTest, VpxImgAllocNv12) { + const int kWidth = 128; + const int kHeight = 128; + + vpx_image_t img; + vpx_img_fmt_t format = VPX_IMG_FMT_NV12; + unsigned int align = 32; + EXPECT_EQ(vpx_img_alloc(&img, format, kWidth, kHeight, align), &img); + EXPECT_EQ(img.stride[VPX_PLANE_U], img.stride[VPX_PLANE_Y]); + EXPECT_EQ(img.stride[VPX_PLANE_V], img.stride[VPX_PLANE_U]); + EXPECT_EQ(img.planes[VPX_PLANE_V], img.planes[VPX_PLANE_U] + 1); + vpx_img_free(&img); +} + +TEST(VpxImageTest, VpxImgAllocHugeWidth) { + // The stride (0x80000000 * 2) would overflow unsigned int. + vpx_image_t *image = + vpx_img_alloc(nullptr, VPX_IMG_FMT_I42016, 0x80000000, 1, 1); + ASSERT_EQ(image, nullptr); + + // The stride (0x80000000) would overflow int. + image = vpx_img_alloc(nullptr, VPX_IMG_FMT_I420, 0x80000000, 1, 1); + ASSERT_EQ(image, nullptr); + + // The aligned width (UINT_MAX + 1) would overflow unsigned int. + image = vpx_img_alloc(nullptr, VPX_IMG_FMT_I420, UINT_MAX, 1, 1); + ASSERT_EQ(image, nullptr); + + image = vpx_img_alloc(nullptr, VPX_IMG_FMT_I420, 0x7ffffffe, 1, 1); + if (image) { + vpx_img_free(image); + } + + image = vpx_img_alloc(nullptr, VPX_IMG_FMT_I420, 285245883, 64, 1); + if (image) { + vpx_img_free(image); + } + + image = vpx_img_alloc(nullptr, VPX_IMG_FMT_NV12, 285245883, 64, 1); + if (image) { + vpx_img_free(image); + } + + image = vpx_img_alloc(nullptr, VPX_IMG_FMT_YV12, 285245883, 64, 1); + if (image) { + vpx_img_free(image); + } + + image = vpx_img_alloc(nullptr, VPX_IMG_FMT_I42016, 65536, 2, 1); + if (image) { + uint16_t *y_plane = + reinterpret_cast<uint16_t *>(image->planes[VPX_PLANE_Y]); + y_plane[0] = 0; + y_plane[image->d_w - 1] = 0; + vpx_img_free(image); + } + + image = vpx_img_alloc(nullptr, VPX_IMG_FMT_I42016, 285245883, 2, 1); + if (image) { + uint16_t *y_plane = + reinterpret_cast<uint16_t *>(image->planes[VPX_PLANE_Y]); + y_plane[0] = 0; + y_plane[image->d_w - 1] = 0; + vpx_img_free(image); + } +} diff --git a/media/libvpx/libvpx/test/yuv_temporal_filter_test.cc b/media/libvpx/libvpx/test/yuv_temporal_filter_test.cc index 0677d55688..5319bfec43 100644 --- a/media/libvpx/libvpx/test/yuv_temporal_filter_test.cc +++ b/media/libvpx/libvpx/test/yuv_temporal_filter_test.cc @@ -14,6 +14,7 @@ #include "test/acm_random.h" #include "test/buffer.h" #include "test/register_state_check.h" +#include "vpx_config.h" #include "vpx_ports/vpx_timer.h" namespace { diff --git a/media/libvpx/libvpx/vp8/common/generic/systemdependent.c b/media/libvpx/libvpx/vp8/common/generic/systemdependent.c index 7c8e083f4f..4aef15eef0 100644 --- a/media/libvpx/libvpx/vp8/common/generic/systemdependent.c +++ b/media/libvpx/libvpx/vp8/common/generic/systemdependent.c @@ -64,19 +64,4 @@ void vp8_machine_specific_config(VP8_COMMON *ctx) { #if CONFIG_MULTITHREAD ctx->processor_core_count = get_cpu_count(); #endif /* CONFIG_MULTITHREAD */ - -#if VPX_ARCH_ARM - ctx->cpu_caps = arm_cpu_caps(); -#elif VPX_ARCH_X86 || VPX_ARCH_X86_64 - ctx->cpu_caps = x86_simd_caps(); -#elif VPX_ARCH_PPC - ctx->cpu_caps = ppc_simd_caps(); -#elif VPX_ARCH_MIPS - ctx->cpu_caps = mips_cpu_caps(); -#elif VPX_ARCH_LOONGARCH - ctx->cpu_caps = loongarch_cpu_caps(); -#else - // generic-gnu targets. - ctx->cpu_caps = 0; -#endif } diff --git a/media/libvpx/libvpx/vp8/common/onyxc_int.h b/media/libvpx/libvpx/vp8/common/onyxc_int.h index ef8d007620..d4824d24e4 100644 --- a/media/libvpx/libvpx/vp8/common/onyxc_int.h +++ b/media/libvpx/libvpx/vp8/common/onyxc_int.h @@ -167,7 +167,6 @@ typedef struct VP8Common { #if CONFIG_POSTPROC struct postproc_state postproc_state; #endif - int cpu_caps; } VP8_COMMON; #ifdef __cplusplus diff --git a/media/libvpx/libvpx/vp8/encoder/bitstream.c b/media/libvpx/libvpx/vp8/encoder/bitstream.c index 03691fc9d1..7bcdf77708 100644 --- a/media/libvpx/libvpx/vp8/encoder/bitstream.c +++ b/media/libvpx/libvpx/vp8/encoder/bitstream.c @@ -501,7 +501,7 @@ static void pack_inter_mode_mvs(VP8_COMP *const cpi) { int ct[4]; vp8_find_near_mvs(xd, m, &n1, &n2, &best_mv, ct, rf, - cpi->common.ref_frame_sign_bias); + pc->ref_frame_sign_bias); vp8_clamp_mv2(&best_mv, xd); vp8_mv_ref_probs(mv_ref_p, ct); @@ -1021,7 +1021,7 @@ void vp8_pack_bitstream(VP8_COMP *cpi, unsigned char *dest, bc[0].error = &pc->error; - validate_buffer(cx_data, 3, cx_data_end, &cpi->common.error); + validate_buffer(cx_data, 3, cx_data_end, &pc->error); cx_data += 3; #if defined(SECTIONBITS_OUTPUT) @@ -1034,7 +1034,7 @@ void vp8_pack_bitstream(VP8_COMP *cpi, unsigned char *dest, if (oh.type == KEY_FRAME) { int v; - validate_buffer(cx_data, 7, cx_data_end, &cpi->common.error); + validate_buffer(cx_data, 7, cx_data_end, &pc->error); /* Start / synch code */ cx_data[0] = 0x9D; @@ -1243,7 +1243,7 @@ void vp8_pack_bitstream(VP8_COMP *cpi, unsigned char *dest, #else if (pc->refresh_entropy_probs == 0) { /* save a copy for later refresh */ - memcpy(&cpi->common.lfc, &cpi->common.fc, sizeof(cpi->common.fc)); + pc->lfc = pc->fc; } vp8_update_coef_probs(cpi); diff --git a/media/libvpx/libvpx/vp8/encoder/encodeframe.c b/media/libvpx/libvpx/vp8/encoder/encodeframe.c index d0117897db..fa43340406 100644 --- a/media/libvpx/libvpx/vp8/encoder/encodeframe.c +++ b/media/libvpx/libvpx/vp8/encoder/encodeframe.c @@ -15,7 +15,6 @@ #include "vp8/common/common.h" #include "vp8/common/entropymode.h" #include "vp8/common/extend.h" -#include "vp8/common/findnearmv.h" #include "vp8/common/invtrans.h" #include "vp8/common/quant_common.h" #include "vp8/common/reconinter.h" @@ -25,11 +24,9 @@ #include "vp8/encoder/encodeframe.h" #include "vp8/encoder/encodeintra.h" #include "vp8/encoder/encodemb.h" -#include "vp8/encoder/encodemv.h" #include "vp8/encoder/onyx_int.h" #include "vp8/encoder/pickinter.h" #include "vp8/encoder/rdopt.h" -#include "vp8/encoder/segmentation.h" #include "vp8_rtcd.h" #include "vpx/internal/vpx_codec_internal.h" #include "vpx_dsp_rtcd.h" diff --git a/media/libvpx/libvpx/vp8/vp8_cx_iface.c b/media/libvpx/libvpx/vp8/vp8_cx_iface.c index 2b238c1a97..668bd0816d 100644 --- a/media/libvpx/libvpx/vp8/vp8_cx_iface.c +++ b/media/libvpx/libvpx/vp8/vp8_cx_iface.c @@ -23,7 +23,6 @@ #include "vpx_mem/vpx_mem.h" #include "vpx_ports/static_assert.h" #include "vpx_ports/system_state.h" -#include "vpx_util/vpx_thread.h" #include "vpx_util/vpx_timestamp.h" #if CONFIG_MULTITHREAD #include "vp8/encoder/ethreading.h" @@ -152,7 +151,7 @@ static vpx_codec_err_t validate_config(vpx_codec_alg_priv_t *ctx, RANGE_CHECK_HI(cfg, g_profile, 3); RANGE_CHECK_HI(cfg, rc_max_quantizer, 63); RANGE_CHECK_HI(cfg, rc_min_quantizer, cfg->rc_max_quantizer); - RANGE_CHECK_HI(cfg, g_threads, MAX_NUM_THREADS); + RANGE_CHECK_HI(cfg, g_threads, 64); #if CONFIG_REALTIME_ONLY RANGE_CHECK_HI(cfg, g_lag_in_frames, 0); #elif CONFIG_MULTI_RES_ENCODING @@ -710,6 +709,7 @@ static vpx_codec_err_t vp8e_init(vpx_codec_ctx_t *ctx, priv->cx_data = malloc(priv->cx_data_sz); if (!priv->cx_data) { + priv->cx_data_sz = 0; return VPX_CODEC_MEM_ERROR; } diff --git a/media/libvpx/libvpx/vp9/decoder/vp9_decodeframe.c b/media/libvpx/libvpx/vp9/decoder/vp9_decodeframe.c index 4fe680cefc..45ef99adf9 100644 --- a/media/libvpx/libvpx/vp9/decoder/vp9_decodeframe.c +++ b/media/libvpx/libvpx/vp9/decoder/vp9_decodeframe.c @@ -68,6 +68,7 @@ static int decode_unsigned_max(struct vpx_read_bit_buffer *rb, int max) { static TX_MODE read_tx_mode(vpx_reader *r) { TX_MODE tx_mode = vpx_read_literal(r, 2); if (tx_mode == ALLOW_32X32) tx_mode += vpx_read_bit(r); + assert(tx_mode < TX_MODES); return tx_mode; } diff --git a/media/libvpx/libvpx/vp9/encoder/vp9_bitstream.c b/media/libvpx/libvpx/vp9/encoder/vp9_bitstream.c index ca56d14aa1..161a6fd535 100644 --- a/media/libvpx/libvpx/vp9/encoder/vp9_bitstream.c +++ b/media/libvpx/libvpx/vp9/encoder/vp9_bitstream.c @@ -9,6 +9,7 @@ */ #include <assert.h> +#include <stdint.h> #include <stdio.h> #include <limits.h> @@ -943,12 +944,11 @@ static int encode_tile_worker(void *arg1, void *arg2) { VP9BitstreamWorkerData *data = (VP9BitstreamWorkerData *)arg2; MACROBLOCKD *const xd = &data->xd; const int tile_row = 0; - vpx_start_encode(&data->bit_writer, data->dest); + vpx_start_encode(&data->bit_writer, data->dest, data->dest_size); write_modes(cpi, xd, &cpi->tile_data[data->tile_idx].tile_info, &data->bit_writer, tile_row, data->tile_idx, &data->max_mv_magnitude, data->interp_filter_selected); - vpx_stop_encode(&data->bit_writer); - return 1; + return vpx_stop_encode(&data->bit_writer) == 0; } void vp9_bitstream_encode_tiles_buffer_dealloc(VP9_COMP *const cpi) { @@ -962,7 +962,18 @@ void vp9_bitstream_encode_tiles_buffer_dealloc(VP9_COMP *const cpi) { } } -static void encode_tiles_buffer_alloc(VP9_COMP *const cpi) { +static size_t encode_tiles_buffer_alloc_size(const VP9_COMP *cpi) { + const VP9_COMMON *cm = &cpi->common; + const int image_bps = + (8 + 2 * (8 >> (cm->subsampling_x + cm->subsampling_y))) * + (1 + (cm->bit_depth > 8)); + const int64_t size = + (int64_t)cpi->oxcf.width * cpi->oxcf.height * image_bps / 8; + return (size_t)size; +} + +static void encode_tiles_buffer_alloc(VP9_COMP *const cpi, + size_t buffer_alloc_size) { VP9_COMMON *const cm = &cpi->common; int i; const size_t worker_data_size = @@ -971,14 +982,14 @@ static void encode_tiles_buffer_alloc(VP9_COMP *const cpi) { vpx_memalign(16, worker_data_size)); memset(cpi->vp9_bitstream_worker_data, 0, worker_data_size); for (i = 1; i < cpi->num_workers; ++i) { - cpi->vp9_bitstream_worker_data[i].dest_size = - cpi->oxcf.width * cpi->oxcf.height; CHECK_MEM_ERROR(&cm->error, cpi->vp9_bitstream_worker_data[i].dest, - vpx_malloc(cpi->vp9_bitstream_worker_data[i].dest_size)); + vpx_malloc(buffer_alloc_size)); + cpi->vp9_bitstream_worker_data[i].dest_size = buffer_alloc_size; } } -static size_t encode_tiles_mt(VP9_COMP *cpi, uint8_t *data_ptr) { +static size_t encode_tiles_mt(VP9_COMP *cpi, uint8_t *data_ptr, + size_t data_size) { const VPxWorkerInterface *const winterface = vpx_get_worker_interface(); VP9_COMMON *const cm = &cpi->common; const int tile_cols = 1 << cm->log2_tile_cols; @@ -986,11 +997,11 @@ static size_t encode_tiles_mt(VP9_COMP *cpi, uint8_t *data_ptr) { size_t total_size = 0; int tile_col = 0; + const size_t buffer_alloc_size = encode_tiles_buffer_alloc_size(cpi); if (!cpi->vp9_bitstream_worker_data || - cpi->vp9_bitstream_worker_data[1].dest_size > - (cpi->oxcf.width * cpi->oxcf.height)) { + cpi->vp9_bitstream_worker_data[1].dest_size != buffer_alloc_size) { vp9_bitstream_encode_tiles_buffer_dealloc(cpi); - encode_tiles_buffer_alloc(cpi); + encode_tiles_buffer_alloc(cpi, buffer_alloc_size); } while (tile_col < tile_cols) { @@ -1010,8 +1021,13 @@ static size_t encode_tiles_mt(VP9_COMP *cpi, uint8_t *data_ptr) { if (i == 0) { // If this worker happens to be for the last tile, then do not offset it // by 4 for the tile size. - data->dest = - data_ptr + total_size + (tile_col == tile_cols - 1 ? 0 : 4); + const size_t offset = total_size + (tile_col == tile_cols - 1 ? 0 : 4); + if (data_size < offset) { + vpx_internal_error(&cm->error, VPX_CODEC_ERROR, + "encode_tiles_mt: output buffer full"); + } + data->dest = data_ptr + offset; + data->dest_size = data_size - offset; } worker->data1 = cpi; worker->data2 = data; @@ -1032,7 +1048,10 @@ static size_t encode_tiles_mt(VP9_COMP *cpi, uint8_t *data_ptr) { uint32_t tile_size; int k; - if (!winterface->sync(worker)) return 0; + if (!winterface->sync(worker)) { + vpx_internal_error(&cm->error, VPX_CODEC_ERROR, + "encode_tiles_mt: worker had error"); + } tile_size = data->bit_writer.pos; // Aggregate per-thread bitstream stats. @@ -1044,10 +1063,18 @@ static size_t encode_tiles_mt(VP9_COMP *cpi, uint8_t *data_ptr) { // Prefix the size of the tile on all but the last. if (tile_col != tile_cols || j < i - 1) { + if (data_size - total_size < 4) { + vpx_internal_error(&cm->error, VPX_CODEC_ERROR, + "encode_tiles_mt: output buffer full"); + } mem_put_be32(data_ptr + total_size, tile_size); total_size += 4; } if (j > 0) { + if (data_size - total_size < tile_size) { + vpx_internal_error(&cm->error, VPX_CODEC_ERROR, + "encode_tiles_mt: output buffer full"); + } memcpy(data_ptr + total_size, data->dest, tile_size); } total_size += tile_size; @@ -1056,7 +1083,7 @@ static size_t encode_tiles_mt(VP9_COMP *cpi, uint8_t *data_ptr) { return total_size; } -static size_t encode_tiles(VP9_COMP *cpi, uint8_t *data_ptr) { +static size_t encode_tiles(VP9_COMP *cpi, uint8_t *data_ptr, size_t data_size) { VP9_COMMON *const cm = &cpi->common; MACROBLOCKD *const xd = &cpi->td.mb.e_mbd; vpx_writer residual_bc; @@ -1073,23 +1100,32 @@ static size_t encode_tiles(VP9_COMP *cpi, uint8_t *data_ptr) { // that it does not make the overall process worse in any case. if (cpi->oxcf.mode == REALTIME && cpi->num_workers > 1 && tile_rows == 1 && tile_cols > 1) { - return encode_tiles_mt(cpi, data_ptr); + return encode_tiles_mt(cpi, data_ptr, data_size); } for (tile_row = 0; tile_row < tile_rows; tile_row++) { for (tile_col = 0; tile_col < tile_cols; tile_col++) { int tile_idx = tile_row * tile_cols + tile_col; + size_t offset; if (tile_col < tile_cols - 1 || tile_row < tile_rows - 1) - vpx_start_encode(&residual_bc, data_ptr + total_size + 4); + offset = total_size + 4; else - vpx_start_encode(&residual_bc, data_ptr + total_size); + offset = total_size; + if (data_size < offset) { + vpx_internal_error(&cm->error, VPX_CODEC_ERROR, + "encode_tiles: output buffer full"); + } + vpx_start_encode(&residual_bc, data_ptr + offset, data_size - offset); write_modes(cpi, xd, &cpi->tile_data[tile_idx].tile_info, &residual_bc, tile_row, tile_col, &cpi->max_mv_magnitude, cpi->interp_filter_selected); - vpx_stop_encode(&residual_bc); + if (vpx_stop_encode(&residual_bc)) { + vpx_internal_error(&cm->error, VPX_CODEC_ERROR, + "encode_tiles: output buffer full"); + } if (tile_col < tile_cols - 1 || tile_row < tile_rows - 1) { // size of this tile mem_put_be32(data_ptr + total_size, residual_bc.pos); @@ -1271,14 +1307,15 @@ static void write_uncompressed_header(VP9_COMP *cpi, write_tile_info(cm, wb); } -static size_t write_compressed_header(VP9_COMP *cpi, uint8_t *data) { +static size_t write_compressed_header(VP9_COMP *cpi, uint8_t *data, + size_t data_size) { VP9_COMMON *const cm = &cpi->common; MACROBLOCKD *const xd = &cpi->td.mb.e_mbd; FRAME_CONTEXT *const fc = cm->fc; FRAME_COUNTS *counts = cpi->td.counts; vpx_writer header_bc; - vpx_start_encode(&header_bc, data); + vpx_start_encode(&header_bc, data, data_size); if (xd->lossless) cm->tx_mode = ONLY_4X4; @@ -1342,46 +1379,68 @@ static size_t write_compressed_header(VP9_COMP *cpi, uint8_t *data) { &counts->mv); } - vpx_stop_encode(&header_bc); - assert(header_bc.pos <= 0xffff); + if (vpx_stop_encode(&header_bc)) { + vpx_internal_error(&cm->error, VPX_CODEC_ERROR, + "write_compressed_header: output buffer full"); + } return header_bc.pos; } -void vp9_pack_bitstream(VP9_COMP *cpi, uint8_t *dest, size_t *size) { +void vp9_pack_bitstream(VP9_COMP *cpi, uint8_t *dest, size_t dest_size, + size_t *size) { + VP9_COMMON *const cm = &cpi->common; uint8_t *data = dest; - size_t first_part_size, uncompressed_hdr_size; - struct vpx_write_bit_buffer wb = { data, 0 }; + size_t data_size = dest_size; + size_t uncompressed_hdr_size, compressed_hdr_size; + struct vpx_write_bit_buffer wb; struct vpx_write_bit_buffer saved_wb; #if CONFIG_BITSTREAM_DEBUG bitstream_queue_reset_write(); #endif + vpx_wb_init(&wb, data, data_size); write_uncompressed_header(cpi, &wb); + if (vpx_wb_has_error(&wb)) { + vpx_internal_error(&cm->error, VPX_CODEC_ERROR, + "vp9_pack_bitstream: output buffer full"); + } // Skip the rest coding process if use show existing frame. - if (cpi->common.show_existing_frame) { + if (cm->show_existing_frame) { uncompressed_hdr_size = vpx_wb_bytes_written(&wb); data += uncompressed_hdr_size; + data_size -= uncompressed_hdr_size; *size = data - dest; return; } saved_wb = wb; - vpx_wb_write_literal(&wb, 0, 16); // don't know in advance first part. size + // don't know in advance compressed header size + vpx_wb_write_literal(&wb, 0, 16); + if (vpx_wb_has_error(&wb)) { + vpx_internal_error(&cm->error, VPX_CODEC_ERROR, + "vp9_pack_bitstream: output buffer full"); + } uncompressed_hdr_size = vpx_wb_bytes_written(&wb); data += uncompressed_hdr_size; + data_size -= uncompressed_hdr_size; vpx_clear_system_state(); - first_part_size = write_compressed_header(cpi, data); - data += first_part_size; - // TODO(jbb): Figure out what to do if first_part_size > 16 bits. - vpx_wb_write_literal(&saved_wb, (int)first_part_size, 16); + compressed_hdr_size = write_compressed_header(cpi, data, data_size); + data += compressed_hdr_size; + data_size -= compressed_hdr_size; + if (compressed_hdr_size > UINT16_MAX) { + vpx_internal_error(&cm->error, VPX_CODEC_ERROR, + "compressed_hdr_size > 16 bits"); + } + vpx_wb_write_literal(&saved_wb, (int)compressed_hdr_size, 16); + assert(!vpx_wb_has_error(&saved_wb)); - data += encode_tiles(cpi, data); + data += encode_tiles(cpi, data, data_size); *size = data - dest; } diff --git a/media/libvpx/libvpx/vp9/encoder/vp9_bitstream.h b/media/libvpx/libvpx/vp9/encoder/vp9_bitstream.h index 208651dc22..1120841ecb 100644 --- a/media/libvpx/libvpx/vp9/encoder/vp9_bitstream.h +++ b/media/libvpx/libvpx/vp9/encoder/vp9_bitstream.h @@ -19,7 +19,7 @@ extern "C" { typedef struct VP9BitstreamWorkerData { uint8_t *dest; - int dest_size; + size_t dest_size; vpx_writer bit_writer; int tile_idx; unsigned int max_mv_magnitude; @@ -35,7 +35,8 @@ int vp9_get_refresh_mask(VP9_COMP *cpi); void vp9_bitstream_encode_tiles_buffer_dealloc(VP9_COMP *const cpi); -void vp9_pack_bitstream(VP9_COMP *cpi, uint8_t *dest, size_t *size); +void vp9_pack_bitstream(VP9_COMP *cpi, uint8_t *dest, size_t dest_size, + size_t *size); static INLINE int vp9_preserve_existing_gf(VP9_COMP *cpi) { return cpi->refresh_golden_frame && cpi->rc.is_src_frame_alt_ref && diff --git a/media/libvpx/libvpx/vp9/encoder/vp9_encodeframe.c b/media/libvpx/libvpx/vp9/encoder/vp9_encodeframe.c index b24c85f406..937f22e835 100644 --- a/media/libvpx/libvpx/vp9/encoder/vp9_encodeframe.c +++ b/media/libvpx/libvpx/vp9/encoder/vp9_encodeframe.c @@ -5853,7 +5853,12 @@ void vp9_init_tile_data(VP9_COMP *cpi) { int tplist_count = 0; if (cpi->tile_data == NULL || cpi->allocated_tiles < tile_cols * tile_rows) { - if (cpi->tile_data != NULL) vpx_free(cpi->tile_data); + if (cpi->tile_data != NULL) { + // Free the row mt memory in cpi->tile_data first. + vp9_row_mt_mem_dealloc(cpi); + vpx_free(cpi->tile_data); + } + cpi->allocated_tiles = 0; CHECK_MEM_ERROR( &cm->error, cpi->tile_data, vpx_malloc(tile_cols * tile_rows * sizeof(*cpi->tile_data))); @@ -5883,9 +5888,9 @@ void vp9_init_tile_data(VP9_COMP *cpi) { for (tile_col = 0; tile_col < tile_cols; ++tile_col) { TileDataEnc *this_tile = &cpi->tile_data[tile_row * tile_cols + tile_col]; TileInfo *tile_info = &this_tile->tile_info; - if (cpi->sf.adaptive_rd_thresh_row_mt && - this_tile->row_base_thresh_freq_fact == NULL) + if (cpi->sf.adaptive_rd_thresh_row_mt) { vp9_row_mt_alloc_rd_thresh(cpi, this_tile); + } vp9_tile_init(tile_info, cm, tile_row, tile_col); cpi->tile_tok[tile_row][tile_col] = pre_tok + tile_tok; diff --git a/media/libvpx/libvpx/vp9/encoder/vp9_encoder.c b/media/libvpx/libvpx/vp9/encoder/vp9_encoder.c index 3b8b5345f1..8616cdc630 100644 --- a/media/libvpx/libvpx/vp9/encoder/vp9_encoder.c +++ b/media/libvpx/libvpx/vp9/encoder/vp9_encoder.c @@ -3901,7 +3901,7 @@ static void set_frame_size(VP9_COMP *cpi) { } if (!frame_is_intra_only(cm) && !has_valid_ref_frame) { vpx_internal_error( - &cm->error, VPX_CODEC_CORRUPT_FRAME, + &cm->error, VPX_CODEC_ERROR, "Can't find at least one reference frame with valid size"); } @@ -3973,7 +3973,7 @@ static YV12_BUFFER_CONFIG *svc_twostage_scale( } static int encode_without_recode_loop(VP9_COMP *cpi, size_t *size, - uint8_t *dest) { + uint8_t *dest, size_t dest_size) { VP9_COMMON *const cm = &cpi->common; SVC *const svc = &cpi->svc; int q = 0, bottom_index = 0, top_index = 0; @@ -4269,7 +4269,7 @@ static int encode_without_recode_loop(VP9_COMP *cpi, size_t *size, int frame_size = 0; // Get an estimate of the encoded frame size. save_coding_context(cpi); - vp9_pack_bitstream(cpi, dest, size); + vp9_pack_bitstream(cpi, dest, dest_size, size); restore_coding_context(cpi); frame_size = (int)(*size) << 3; // Check if encoded frame will overshoot too much, and if so, set the q and @@ -4472,7 +4472,8 @@ static void rq_model_update(const RATE_QINDEX_HISTORY *rq_history, } #endif // CONFIG_RATE_CTRL -static void encode_with_recode_loop(VP9_COMP *cpi, size_t *size, uint8_t *dest +static void encode_with_recode_loop(VP9_COMP *cpi, size_t *size, uint8_t *dest, + size_t dest_size #if CONFIG_RATE_CTRL , RATE_QINDEX_HISTORY *rq_history @@ -4665,7 +4666,8 @@ static void encode_with_recode_loop(VP9_COMP *cpi, size_t *size, uint8_t *dest // to recode. if (cpi->sf.recode_loop >= ALLOW_RECODE_KFARFGF) { save_coding_context(cpi); - if (!cpi->sf.use_nonrd_pick_mode) vp9_pack_bitstream(cpi, dest, size); + if (!cpi->sf.use_nonrd_pick_mode) + vp9_pack_bitstream(cpi, dest, dest_size, size); rc->projected_frame_size = (int)(*size) << 3; @@ -5173,7 +5175,7 @@ static void spatial_denoise_frame(VP9_COMP *cpi) { #if !CONFIG_REALTIME_ONLY static void vp9_try_disable_lookahead_aq(VP9_COMP *cpi, size_t *size, - uint8_t *dest) { + uint8_t *dest, size_t dest_size) { if (cpi->common.seg.enabled) if (ALT_REF_AQ_PROTECT_GAIN) { size_t nsize = *size; @@ -5184,7 +5186,7 @@ static void vp9_try_disable_lookahead_aq(VP9_COMP *cpi, size_t *size, save_coding_context(cpi); vp9_disable_segmentation(&cpi->common.seg); - vp9_pack_bitstream(cpi, dest, &nsize); + vp9_pack_bitstream(cpi, dest, dest_size, &nsize); restore_coding_context(cpi); overhead = (int)*size - (int)nsize; @@ -5477,8 +5479,8 @@ static void update_encode_frame_result_simple_encode( #endif // !CONFIG_REALTIME_ONLY static void encode_frame_to_data_rate( - VP9_COMP *cpi, size_t *size, uint8_t *dest, unsigned int *frame_flags, - ENCODE_FRAME_RESULT *encode_frame_result) { + VP9_COMP *cpi, size_t *size, uint8_t *dest, size_t dest_size, + unsigned int *frame_flags, ENCODE_FRAME_RESULT *encode_frame_result) { VP9_COMMON *const cm = &cpi->common; const VP9EncoderConfig *const oxcf = &cpi->oxcf; struct segmentation *const seg = &cm->seg; @@ -5585,16 +5587,17 @@ static void encode_frame_to_data_rate( } if (cpi->sf.recode_loop == DISALLOW_RECODE) { - if (!encode_without_recode_loop(cpi, size, dest)) return; + if (!encode_without_recode_loop(cpi, size, dest, dest_size)) return; } else { #if !CONFIG_REALTIME_ONLY #if CONFIG_RATE_CTRL - encode_with_recode_loop(cpi, size, dest, &encode_frame_result->rq_history); + encode_with_recode_loop(cpi, size, dest, dest_size, + &encode_frame_result->rq_history); #else // CONFIG_RATE_CTRL #if CONFIG_COLLECT_COMPONENT_TIMING start_timing(cpi, encode_with_recode_loop_time); #endif - encode_with_recode_loop(cpi, size, dest); + encode_with_recode_loop(cpi, size, dest, dest_size); #if CONFIG_COLLECT_COMPONENT_TIMING end_timing(cpi, encode_with_recode_loop_time); #endif @@ -5614,7 +5617,7 @@ static void encode_frame_to_data_rate( #if !CONFIG_REALTIME_ONLY // Disable segmentation if it decrease rate/distortion ratio if (cpi->oxcf.aq_mode == LOOKAHEAD_AQ) - vp9_try_disable_lookahead_aq(cpi, size, dest); + vp9_try_disable_lookahead_aq(cpi, size, dest, dest_size); #endif #if CONFIG_VP9_TEMPORAL_DENOISING @@ -5671,7 +5674,7 @@ static void encode_frame_to_data_rate( start_timing(cpi, vp9_pack_bitstream_time); #endif // build the bitstream - vp9_pack_bitstream(cpi, dest, size); + vp9_pack_bitstream(cpi, dest, dest_size, size); #if CONFIG_COLLECT_COMPONENT_TIMING end_timing(cpi, vp9_pack_bitstream_time); #endif @@ -5862,32 +5865,33 @@ static void encode_frame_to_data_rate( } static void SvcEncode(VP9_COMP *cpi, size_t *size, uint8_t *dest, - unsigned int *frame_flags) { + size_t dest_size, unsigned int *frame_flags) { vp9_rc_get_svc_params(cpi); - encode_frame_to_data_rate(cpi, size, dest, frame_flags, + encode_frame_to_data_rate(cpi, size, dest, dest_size, frame_flags, /*encode_frame_result = */ NULL); } static void Pass0Encode(VP9_COMP *cpi, size_t *size, uint8_t *dest, - unsigned int *frame_flags) { + size_t dest_size, unsigned int *frame_flags) { if (cpi->oxcf.rc_mode == VPX_CBR) { vp9_rc_get_one_pass_cbr_params(cpi); } else { vp9_rc_get_one_pass_vbr_params(cpi); } - encode_frame_to_data_rate(cpi, size, dest, frame_flags, + encode_frame_to_data_rate(cpi, size, dest, dest_size, frame_flags, /*encode_frame_result = */ NULL); } #if !CONFIG_REALTIME_ONLY static void Pass2Encode(VP9_COMP *cpi, size_t *size, uint8_t *dest, - unsigned int *frame_flags, + size_t dest_size, unsigned int *frame_flags, ENCODE_FRAME_RESULT *encode_frame_result) { cpi->allow_encode_breakout = ENCODE_BREAKOUT_ENABLED; #if CONFIG_MISMATCH_DEBUG mismatch_move_frame_idx_w(); #endif - encode_frame_to_data_rate(cpi, size, dest, frame_flags, encode_frame_result); + encode_frame_to_data_rate(cpi, size, dest, dest_size, frame_flags, + encode_frame_result); } #endif // !CONFIG_REALTIME_ONLY @@ -6300,8 +6304,8 @@ void vp9_init_encode_frame_result(ENCODE_FRAME_RESULT *encode_frame_result) { } int vp9_get_compressed_data(VP9_COMP *cpi, unsigned int *frame_flags, - size_t *size, uint8_t *dest, int64_t *time_stamp, - int64_t *time_end, int flush, + size_t *size, uint8_t *dest, size_t dest_size, + int64_t *time_stamp, int64_t *time_end, int flush, ENCODE_FRAME_RESULT *encode_frame_result) { const VP9EncoderConfig *const oxcf = &cpi->oxcf; VP9_COMMON *const cm = &cpi->common; @@ -6583,10 +6587,10 @@ int vp9_get_compressed_data(VP9_COMP *cpi, unsigned int *frame_flags, #if CONFIG_REALTIME_ONLY (void)encode_frame_result; if (cpi->use_svc) { - SvcEncode(cpi, size, dest, frame_flags); + SvcEncode(cpi, size, dest, dest_size, frame_flags); } else { // One pass encode - Pass0Encode(cpi, size, dest, frame_flags); + Pass0Encode(cpi, size, dest, dest_size, frame_flags); } #else // !CONFIG_REALTIME_ONLY if (oxcf->pass == 1 && !cpi->use_svc) { @@ -6609,16 +6613,16 @@ int vp9_get_compressed_data(VP9_COMP *cpi, unsigned int *frame_flags, // Accumulate 2nd pass time in 2-pass case. start_timing(cpi, Pass2Encode_time); #endif - Pass2Encode(cpi, size, dest, frame_flags, encode_frame_result); + Pass2Encode(cpi, size, dest, dest_size, frame_flags, encode_frame_result); vp9_twopass_postencode_update(cpi); #if CONFIG_COLLECT_COMPONENT_TIMING end_timing(cpi, Pass2Encode_time); #endif } else if (cpi->use_svc) { - SvcEncode(cpi, size, dest, frame_flags); + SvcEncode(cpi, size, dest, dest_size, frame_flags); } else { // One pass encode - Pass0Encode(cpi, size, dest, frame_flags); + Pass0Encode(cpi, size, dest, dest_size, frame_flags); } #endif // CONFIG_REALTIME_ONLY diff --git a/media/libvpx/libvpx/vp9/encoder/vp9_encoder.h b/media/libvpx/libvpx/vp9/encoder/vp9_encoder.h index 898855d10d..e3e740f17e 100644 --- a/media/libvpx/libvpx/vp9/encoder/vp9_encoder.h +++ b/media/libvpx/libvpx/vp9/encoder/vp9_encoder.h @@ -339,6 +339,10 @@ typedef struct TileDataEnc { // Used for adaptive_rd_thresh with row multithreading int *row_base_thresh_freq_fact; + // The value of sb_rows when row_base_thresh_freq_fact is allocated. + // The row_base_thresh_freq_fact array has sb_rows * BLOCK_SIZES * MAX_MODES + // elements. + int sb_rows; MV firstpass_top_mv; } TileDataEnc; @@ -1221,8 +1225,8 @@ int vp9_receive_raw_frame(VP9_COMP *cpi, vpx_enc_frame_flags_t frame_flags, int64_t end_time); int vp9_get_compressed_data(VP9_COMP *cpi, unsigned int *frame_flags, - size_t *size, uint8_t *dest, int64_t *time_stamp, - int64_t *time_end, int flush, + size_t *size, uint8_t *dest, size_t dest_size, + int64_t *time_stamp, int64_t *time_end, int flush, ENCODE_FRAME_RESULT *encode_frame_result); int vp9_get_preview_raw_frame(VP9_COMP *cpi, YV12_BUFFER_CONFIG *dest, diff --git a/media/libvpx/libvpx/vp9/encoder/vp9_ethread.h b/media/libvpx/libvpx/vp9/encoder/vp9_ethread.h index 359cdd1290..46478bef9f 100644 --- a/media/libvpx/libvpx/vp9/encoder/vp9_ethread.h +++ b/media/libvpx/libvpx/vp9/encoder/vp9_ethread.h @@ -19,6 +19,7 @@ extern "C" { #define MAX_NUM_TILE_COLS (1 << 6) #define MAX_NUM_TILE_ROWS 4 +#define MAX_NUM_THREADS 64 struct VP9_COMP; struct ThreadData; diff --git a/media/libvpx/libvpx/vp9/encoder/vp9_firstpass.c b/media/libvpx/libvpx/vp9/encoder/vp9_firstpass.c index 58b9b7ba61..2bf1bfcaf4 100644 --- a/media/libvpx/libvpx/vp9/encoder/vp9_firstpass.c +++ b/media/libvpx/libvpx/vp9/encoder/vp9_firstpass.c @@ -2302,6 +2302,44 @@ static void define_gf_group_structure(VP9_COMP *cpi) { gf_group->gf_group_size = frame_index; } +static void ext_rc_define_gf_group_structure( + VP9_COMP *cpi, vpx_rc_gop_decision_t *gop_decision) { + RATE_CONTROL *const rc = &cpi->rc; + TWO_PASS *const twopass = &cpi->twopass; + GF_GROUP *const gf_group = &twopass->gf_group; + const int key_frame = cpi->common.frame_type == KEY_FRAME; + + if (!key_frame) { + set_gf_overlay_frame_type(gf_group, 0, rc->source_alt_ref_active); + } + + for (int frame_index = 1; frame_index < gop_decision->gop_coding_frames; + frame_index++) { + const int ext_frame_index = key_frame ? frame_index : frame_index - 1; + const vpx_rc_frame_update_type_t update_type = + gop_decision->update_type[ext_frame_index]; + gf_group->update_type[frame_index] = (FRAME_UPDATE_TYPE)update_type; + if (update_type == VPX_RC_ARF_UPDATE) { + gf_group->rf_level[frame_index] = GF_ARF_STD; + gf_group->layer_depth[frame_index] = 1; + gf_group->arf_src_offset[frame_index] = + (unsigned char)(rc->baseline_gf_interval - 1); + gf_group->frame_gop_index[frame_index] = rc->baseline_gf_interval; + } else if (update_type == VPX_RC_LF_UPDATE) { + gf_group->frame_gop_index[frame_index] = frame_index; + gf_group->arf_src_offset[frame_index] = 0; + gf_group->rf_level[frame_index] = INTER_NORMAL; + gf_group->layer_depth[frame_index] = 2; + } else if (update_type == VPX_RC_OVERLAY_UPDATE) { + set_gf_overlay_frame_type(gf_group, frame_index, + rc->source_alt_ref_pending); + gf_group->arf_src_offset[frame_index] = 0; + gf_group->frame_gop_index[frame_index] = rc->baseline_gf_interval; + } + } + gf_group->max_layer_depth = 2; +} + static void allocate_gf_group_bits(VP9_COMP *cpi, int64_t gf_group_bits, int gf_arf_bits) { VP9EncoderConfig *const oxcf = &cpi->oxcf; @@ -3604,7 +3642,7 @@ void vp9_rc_get_second_pass_params(VP9_COMP *cpi) { rc->baseline_gf_interval = gop_decision.gop_coding_frames - rc->source_alt_ref_pending; rc->frames_till_gf_update_due = rc->baseline_gf_interval; - define_gf_group_structure(cpi); + ext_rc_define_gf_group_structure(cpi, &gop_decision); } } else { // Keyframe and section processing. diff --git a/media/libvpx/libvpx/vp9/encoder/vp9_multi_thread.c b/media/libvpx/libvpx/vp9/encoder/vp9_multi_thread.c index 6e124f9944..8437ce7531 100644 --- a/media/libvpx/libvpx/vp9/encoder/vp9_multi_thread.c +++ b/media/libvpx/libvpx/vp9/encoder/vp9_multi_thread.c @@ -55,16 +55,23 @@ void *vp9_enc_grp_get_next_job(MultiThreadHandle *multi_thread_ctxt, void vp9_row_mt_alloc_rd_thresh(VP9_COMP *const cpi, TileDataEnc *const this_tile) { VP9_COMMON *const cm = &cpi->common; - const int sb_rows = - (mi_cols_aligned_to_sb(cm->mi_rows) >> MI_BLOCK_SIZE_LOG2) + 1; + const int sb_rows = mi_cols_aligned_to_sb(cm->mi_rows) >> MI_BLOCK_SIZE_LOG2; int i; + if (this_tile->row_base_thresh_freq_fact != NULL) { + if (sb_rows <= this_tile->sb_rows) { + return; + } + vpx_free(this_tile->row_base_thresh_freq_fact); + this_tile->row_base_thresh_freq_fact = NULL; + } CHECK_MEM_ERROR( &cm->error, this_tile->row_base_thresh_freq_fact, (int *)vpx_calloc(sb_rows * BLOCK_SIZES * MAX_MODES, sizeof(*(this_tile->row_base_thresh_freq_fact)))); for (i = 0; i < sb_rows * BLOCK_SIZES * MAX_MODES; i++) this_tile->row_base_thresh_freq_fact[i] = RD_THRESH_INIT_FACT; + this_tile->sb_rows = sb_rows; } void vp9_row_mt_mem_alloc(VP9_COMP *cpi) { @@ -101,13 +108,6 @@ void vp9_row_mt_mem_alloc(VP9_COMP *cpi) { for (tile_col = 0; tile_col < tile_cols; tile_col++) { TileDataEnc *this_tile = &cpi->tile_data[tile_col]; vp9_row_mt_sync_mem_alloc(&this_tile->row_mt_sync, cm, jobs_per_tile_col); - if (cpi->sf.adaptive_rd_thresh_row_mt) { - if (this_tile->row_base_thresh_freq_fact != NULL) { - vpx_free(this_tile->row_base_thresh_freq_fact); - this_tile->row_base_thresh_freq_fact = NULL; - } - vp9_row_mt_alloc_rd_thresh(cpi, this_tile); - } } // Assign the sync pointer of tile row zero for every tile row > 0 @@ -136,14 +136,17 @@ void vp9_row_mt_mem_dealloc(VP9_COMP *cpi) { #endif // Deallocate memory for job queue - if (multi_thread_ctxt->job_queue) vpx_free(multi_thread_ctxt->job_queue); + if (multi_thread_ctxt->job_queue) { + vpx_free(multi_thread_ctxt->job_queue); + multi_thread_ctxt->job_queue = NULL; + } #if CONFIG_MULTITHREAD // Destroy mutex for each tile for (tile_col = 0; tile_col < multi_thread_ctxt->allocated_tile_cols; tile_col++) { RowMTInfo *row_mt_info = &multi_thread_ctxt->row_mt_info[tile_col]; - if (row_mt_info) pthread_mutex_destroy(&row_mt_info->job_mutex); + pthread_mutex_destroy(&row_mt_info->job_mutex); } #endif @@ -169,6 +172,10 @@ void vp9_row_mt_mem_dealloc(VP9_COMP *cpi) { } } #endif + + multi_thread_ctxt->allocated_tile_cols = 0; + multi_thread_ctxt->allocated_tile_rows = 0; + multi_thread_ctxt->allocated_vert_unit_rows = 0; } void vp9_multi_thread_tile_init(VP9_COMP *cpi) { diff --git a/media/libvpx/libvpx/vp9/encoder/vp9_pickmode.c b/media/libvpx/libvpx/vp9/encoder/vp9_pickmode.c index 6f2524b36e..d561b5062c 100644 --- a/media/libvpx/libvpx/vp9/encoder/vp9_pickmode.c +++ b/media/libvpx/libvpx/vp9/encoder/vp9_pickmode.c @@ -1698,7 +1698,7 @@ void vp9_pick_inter_mode(VP9_COMP *cpi, MACROBLOCK *x, TileDataEnc *tile_data, MV_REFERENCE_FRAME usable_ref_frame, second_ref_frame; int_mv frame_mv[MB_MODE_COUNT][MAX_REF_FRAMES]; uint8_t mode_checked[MB_MODE_COUNT][MAX_REF_FRAMES]; - struct buf_2d yv12_mb[4][MAX_MB_PLANE]; + struct buf_2d yv12_mb[4][MAX_MB_PLANE] = { 0 }; RD_COST this_rdc, best_rdc; // var_y and sse_y are saved to be used in skipping checking unsigned int var_y = UINT_MAX; diff --git a/media/libvpx/libvpx/vp9/encoder/vp9_rdopt.c b/media/libvpx/libvpx/vp9/encoder/vp9_rdopt.c index 447136ed84..c1a8c10891 100644 --- a/media/libvpx/libvpx/vp9/encoder/vp9_rdopt.c +++ b/media/libvpx/libvpx/vp9/encoder/vp9_rdopt.c @@ -3435,6 +3435,14 @@ int vp9_active_edge_sb(VP9_COMP *cpi, int mi_row, int mi_col) { } #if !CONFIG_REALTIME_ONLY +void init_frame_mv(int_mv frame_mv[MB_MODE_COUNT][MAX_REF_FRAMES]) { + for (int mode = 0; mode < MB_MODE_COUNT; ++mode) { + for (int ref_frame = 0; ref_frame < MAX_REF_FRAMES; ++ref_frame) { + frame_mv[mode][ref_frame].as_int = INVALID_MV; + } + } +} + void vp9_rd_pick_inter_mode_sb(VP9_COMP *cpi, TileDataEnc *tile_data, MACROBLOCK *x, int mi_row, int mi_col, RD_COST *rd_cost, BLOCK_SIZE bsize, @@ -3452,7 +3460,7 @@ void vp9_rd_pick_inter_mode_sb(VP9_COMP *cpi, TileDataEnc *tile_data, unsigned char segment_id = mi->segment_id; int comp_pred, i, k; int_mv frame_mv[MB_MODE_COUNT][MAX_REF_FRAMES]; - struct buf_2d yv12_mb[4][MAX_MB_PLANE]; + struct buf_2d yv12_mb[4][MAX_MB_PLANE] = { 0 }; int_mv single_newmv[MAX_REF_FRAMES] = { { 0 } }; INTERP_FILTER single_inter_filter[MB_MODE_COUNT][MAX_REF_FRAMES]; int single_skippable[MB_MODE_COUNT][MAX_REF_FRAMES]; @@ -3530,6 +3538,8 @@ void vp9_rd_pick_inter_mode_sb(VP9_COMP *cpi, TileDataEnc *tile_data, rd_cost->rate = INT_MAX; + init_frame_mv(frame_mv); + for (ref_frame = LAST_FRAME; ref_frame <= ALTREF_FRAME; ++ref_frame) { x->pred_mv_sad[ref_frame] = INT_MAX; if ((cpi->ref_frame_flags & ref_frame_to_flag(ref_frame)) && @@ -4297,7 +4307,7 @@ void vp9_rd_pick_inter_mode_sub8x8(VP9_COMP *cpi, TileDataEnc *tile_data, unsigned char segment_id = mi->segment_id; int comp_pred, i; int_mv frame_mv[MB_MODE_COUNT][MAX_REF_FRAMES]; - struct buf_2d yv12_mb[4][MAX_MB_PLANE]; + struct buf_2d yv12_mb[4][MAX_MB_PLANE] = { 0 }; int64_t best_rd = best_rd_so_far; int64_t best_yrd = best_rd_so_far; // FIXME(rbultje) more precise int64_t best_pred_diff[REFERENCE_MODES]; diff --git a/media/libvpx/libvpx/vp9/encoder/vp9_speed_features.h b/media/libvpx/libvpx/vp9/encoder/vp9_speed_features.h index 941de639ac..513ae93526 100644 --- a/media/libvpx/libvpx/vp9/encoder/vp9_speed_features.h +++ b/media/libvpx/libvpx/vp9/encoder/vp9_speed_features.h @@ -306,7 +306,7 @@ typedef struct SPEED_FEATURES { // Turned off when (row_mt_bit_exact == 1 && adaptive_rd_thresh_row_mt == 0). int adaptive_rd_thresh; - // Flag to use adaptive_rd_thresh when row-mt it enabled, only for non-rd + // Flag to use adaptive_rd_thresh when row-mt is enabled, only for non-rd // pickmode. int adaptive_rd_thresh_row_mt; diff --git a/media/libvpx/libvpx/vp9/ratectrl_rtc.cc b/media/libvpx/libvpx/vp9/ratectrl_rtc.cc index 942c15ce49..a1be595353 100644 --- a/media/libvpx/libvpx/vp9/ratectrl_rtc.cc +++ b/media/libvpx/libvpx/vp9/ratectrl_rtc.cc @@ -305,7 +305,9 @@ int VP9RateControlRTC::GetLoopfilterLevel() const { bool VP9RateControlRTC::GetSegmentationData( VP9SegmentationData *segmentation_data) const { - if (!cpi_->cyclic_refresh->apply_cyclic_refresh) return false; + if (!cpi_->cyclic_refresh || !cpi_->cyclic_refresh->apply_cyclic_refresh) { + return false; + } segmentation_data->segmentation_map = cpi_->segmentation_map; segmentation_data->segmentation_map_size = diff --git a/media/libvpx/libvpx/vp9/simple_encode.cc b/media/libvpx/libvpx/vp9/simple_encode.cc index 5e565d1b1a..54b4f38559 100644 --- a/media/libvpx/libvpx/vp9/simple_encode.cc +++ b/media/libvpx/libvpx/vp9/simple_encode.cc @@ -502,6 +502,7 @@ static bool init_encode_frame_result(EncodeFrameResult *encode_frame_result, encode_frame_result->coding_data.reset( new (std::nothrow) uint8_t[max_coding_data_byte_size]); + encode_frame_result->max_coding_data_byte_size = max_coding_data_byte_size; encode_frame_result->num_rows_4x4 = get_num_unit_4x4(frame_height); encode_frame_result->num_cols_4x4 = get_num_unit_4x4(frame_width); @@ -512,6 +513,7 @@ static bool init_encode_frame_result(EncodeFrameResult *encode_frame_result, encode_frame_result->tpl_stats_info.resize(MAX_LAG_BUFFERS); if (encode_frame_result->coding_data.get() == nullptr) { + encode_frame_result->max_coding_data_byte_size = 0; return false; } return init_image_buffer(&encode_frame_result->coded_frame, frame_width, @@ -919,7 +921,7 @@ void SimpleEncode::ComputeFirstPassStats() { ENCODE_FRAME_RESULT encode_frame_info; vp9_init_encode_frame_result(&encode_frame_info); // TODO(angiebird): Call vp9_first_pass directly - vp9_get_compressed_data(impl_ptr_->cpi, &frame_flags, &size, nullptr, + vp9_get_compressed_data(impl_ptr_->cpi, &frame_flags, &size, nullptr, 0, &time_stamp, &time_end, flush, &encode_frame_info); // vp9_get_compressed_data only generates first pass stats not @@ -1205,8 +1207,9 @@ void SimpleEncode::EncodeFrame(EncodeFrameResult *encode_frame_result) { &encode_frame_info.coded_frame); vp9_get_compressed_data(cpi, &frame_flags, &encode_frame_result->coding_data_byte_size, - encode_frame_result->coding_data.get(), &time_stamp, - &time_end, flush, &encode_frame_info); + encode_frame_result->coding_data.get(), + encode_frame_result->max_coding_data_byte_size, + &time_stamp, &time_end, flush, &encode_frame_info); if (out_file_ != nullptr) { ivf_write_frame_header(out_file_, time_stamp, encode_frame_result->coding_data_byte_size); @@ -1220,10 +1223,8 @@ void SimpleEncode::EncodeFrame(EncodeFrameResult *encode_frame_result) { fprintf(stderr, "Coding data size <= 0\n"); abort(); } - const size_t max_coding_data_byte_size = - get_max_coding_data_byte_size(frame_width_, frame_height_); if (encode_frame_result->coding_data_byte_size > - max_coding_data_byte_size) { + encode_frame_result->max_coding_data_byte_size) { fprintf(stderr, "Coding data size exceeds the maximum.\n"); abort(); } diff --git a/media/libvpx/libvpx/vp9/simple_encode.h b/media/libvpx/libvpx/vp9/simple_encode.h index d610a5e159..94ecbf284c 100644 --- a/media/libvpx/libvpx/vp9/simple_encode.h +++ b/media/libvpx/libvpx/vp9/simple_encode.h @@ -263,6 +263,7 @@ struct EncodeFrameResult { // The EncodeFrame will allocate a buffer, write the coding data into the // buffer and give the ownership of the buffer to coding_data. std::unique_ptr<unsigned char[]> coding_data; + size_t max_coding_data_byte_size; double psnr; uint64_t sse; int quantize_index; diff --git a/media/libvpx/libvpx/vp9/vp9_cx_iface.c b/media/libvpx/libvpx/vp9/vp9_cx_iface.c index fe62bac5f2..f365e578d8 100644 --- a/media/libvpx/libvpx/vp9/vp9_cx_iface.c +++ b/media/libvpx/libvpx/vp9/vp9_cx_iface.c @@ -19,11 +19,11 @@ #include "vpx_dsp/psnr.h" #include "vpx_ports/static_assert.h" #include "vpx_ports/system_state.h" -#include "vpx_util/vpx_thread.h" #include "vpx_util/vpx_timestamp.h" #include "vpx/internal/vpx_codec_internal.h" #include "./vpx_version.h" #include "vp9/encoder/vp9_encoder.h" +#include "vp9/encoder/vp9_ethread.h" #include "vpx/vp8cx.h" #include "vp9/common/vp9_alloccommon.h" #include "vp9/common/vp9_scale.h" @@ -1459,13 +1459,14 @@ static vpx_codec_err_t encoder_encode(vpx_codec_alg_priv_t *ctx, /* Any pending invisible frames? */ if (ctx->pending_cx_data) { + assert(cx_data_sz >= ctx->pending_cx_data_sz); memmove(cx_data, ctx->pending_cx_data, ctx->pending_cx_data_sz); ctx->pending_cx_data = cx_data; cx_data += ctx->pending_cx_data_sz; cx_data_sz -= ctx->pending_cx_data_sz; - /* TODO: this is a minimal check, the underlying codec doesn't respect - * the buffer size anyway. + /* TODO(webm:1844): this is a minimal check, the underlying codec doesn't + * respect the buffer size anyway. */ if (cx_data_sz < ctx->cx_data_sz / 2) { vpx_internal_error(&cpi->common.error, VPX_CODEC_ERROR, @@ -1484,9 +1485,9 @@ static vpx_codec_err_t encoder_encode(vpx_codec_alg_priv_t *ctx, ENCODE_FRAME_RESULT encode_frame_result; vp9_init_encode_frame_result(&encode_frame_result); // TODO(angiebird): Call vp9_first_pass directly - ret = vp9_get_compressed_data(cpi, &lib_flags, &size, cx_data, - &dst_time_stamp, &dst_end_time_stamp, - !img, &encode_frame_result); + ret = vp9_get_compressed_data( + cpi, &lib_flags, &size, cx_data, cx_data_sz, &dst_time_stamp, + &dst_end_time_stamp, !img, &encode_frame_result); assert(size == 0); // There is no compressed data in the first pass (void)ret; assert(ret == 0); @@ -1510,8 +1511,9 @@ static vpx_codec_err_t encoder_encode(vpx_codec_alg_priv_t *ctx, vp9_init_encode_frame_result(&encode_frame_result); while (cx_data_sz >= ctx->cx_data_sz / 2 && -1 != vp9_get_compressed_data(cpi, &lib_flags, &size, cx_data, - &dst_time_stamp, &dst_end_time_stamp, - !img, &encode_frame_result)) { + cx_data_sz, &dst_time_stamp, + &dst_end_time_stamp, !img, + &encode_frame_result)) { // Pack psnr pkt if (size > 0 && !cpi->use_svc) { // TODO(angiebird): Figure out while we don't need psnr pkt when @@ -1528,7 +1530,7 @@ static vpx_codec_err_t encoder_encode(vpx_codec_alg_priv_t *ctx, if (!cpi->common.show_frame || (cpi->use_svc && cpi->svc.spatial_layer_id < cpi->svc.number_spatial_layers - 1)) { - if (ctx->pending_cx_data == 0) ctx->pending_cx_data = cx_data; + if (ctx->pending_cx_data == NULL) ctx->pending_cx_data = cx_data; ctx->pending_cx_data_sz += size; if (size) ctx->pending_frame_sizes[ctx->pending_frame_count++] = size; diff --git a/media/libvpx/libvpx/vpx/exports_com b/media/libvpx/libvpx/vpx/exports_com index f0b46aa175..2ab05099f8 100644 --- a/media/libvpx/libvpx/vpx/exports_com +++ b/media/libvpx/libvpx/vpx/exports_com @@ -14,6 +14,3 @@ text vpx_img_flip text vpx_img_free text vpx_img_set_rect text vpx_img_wrap -text vpx_free_tpl_gop_stats -text vpx_read_tpl_gop_stats -text vpx_write_tpl_gop_stats diff --git a/media/libvpx/libvpx/vpx/src/vpx_image.c b/media/libvpx/libvpx/vpx/src/vpx_image.c index 3f7ff74244..8ef218a252 100644 --- a/media/libvpx/libvpx/vpx/src/vpx_image.c +++ b/media/libvpx/libvpx/vpx/src/vpx_image.c @@ -8,6 +8,7 @@ * be found in the AUTHORS file in the root of the source tree. */ +#include <assert.h> #include <limits.h> #include <stdlib.h> #include <string.h> @@ -21,14 +22,23 @@ static vpx_image_t *img_alloc_helper(vpx_image_t *img, vpx_img_fmt_t fmt, unsigned int buf_align, unsigned int stride_align, unsigned char *img_data) { - unsigned int h, w, s, xcs, ycs, bps; - unsigned int stride_in_bytes; + unsigned int h, w, xcs, ycs, bps; + uint64_t s; + int stride_in_bytes; unsigned int align; if (img != NULL) memset(img, 0, sizeof(vpx_image_t)); if (fmt == VPX_IMG_FMT_NONE) goto fail; + /* Impose maximum values on input parameters so that this function can + * perform arithmetic operations without worrying about overflows. + */ + if (d_w > 0x08000000 || d_h > 0x08000000 || buf_align > 65536 || + stride_align > 65536) { + goto fail; + } + /* Treat align==0 like align==1 */ if (!buf_align) buf_align = 1; @@ -78,13 +88,28 @@ static vpx_image_t *img_alloc_helper(vpx_image_t *img, vpx_img_fmt_t fmt, default: ycs = 0; break; } - /* Calculate storage sizes. If the buffer was allocated externally, the width - * and height shouldn't be adjusted. */ - w = d_w; - h = d_h; - s = (fmt & VPX_IMG_FMT_PLANAR) ? w : bps * w / 8; - s = (s + stride_align - 1) & ~(stride_align - 1); - stride_in_bytes = (fmt & VPX_IMG_FMT_HIGHBITDEPTH) ? s * 2 : s; + /* Calculate storage sizes. */ + if (img_data) { + /* If the buffer was allocated externally, the width and height shouldn't + * be adjusted. */ + w = d_w; + h = d_h; + } else { + /* Calculate storage sizes given the chroma subsampling */ + align = (1 << xcs) - 1; + w = (d_w + align) & ~align; + assert(d_w <= w); + align = (1 << ycs) - 1; + h = (d_h + align) & ~align; + assert(d_h <= h); + } + + s = (fmt & VPX_IMG_FMT_PLANAR) ? w : (uint64_t)bps * w / 8; + s = (fmt & VPX_IMG_FMT_HIGHBITDEPTH) ? s * 2 : s; + s = (s + stride_align - 1) & ~((uint64_t)stride_align - 1); + if (s > INT_MAX) goto fail; + stride_in_bytes = (int)s; + s = (fmt & VPX_IMG_FMT_HIGHBITDEPTH) ? s / 2 : s; /* Allocate the new image */ if (!img) { @@ -99,15 +124,6 @@ static vpx_image_t *img_alloc_helper(vpx_image_t *img, vpx_img_fmt_t fmt, if (!img_data) { uint64_t alloc_size; - /* Calculate storage sizes given the chroma subsampling */ - align = (1 << xcs) - 1; - w = (d_w + align) & ~align; - align = (1 << ycs) - 1; - h = (d_h + align) & ~align; - - s = (fmt & VPX_IMG_FMT_PLANAR) ? w : bps * w / 8; - s = (s + stride_align - 1) & ~(stride_align - 1); - stride_in_bytes = (fmt & VPX_IMG_FMT_HIGHBITDEPTH) ? s * 2 : s; alloc_size = (fmt & VPX_IMG_FMT_PLANAR) ? (uint64_t)h * s * bps / 8 : (uint64_t)h * s; @@ -148,8 +164,8 @@ vpx_image_t *vpx_img_alloc(vpx_image_t *img, vpx_img_fmt_t fmt, vpx_image_t *vpx_img_wrap(vpx_image_t *img, vpx_img_fmt_t fmt, unsigned int d_w, unsigned int d_h, unsigned int stride_align, unsigned char *img_data) { - /* By setting buf_align = 1, we don't change buffer alignment in this - * function. */ + /* Set buf_align = 1. It is ignored by img_alloc_helper because img_data is + * not NULL. */ return img_alloc_helper(img, fmt, d_w, d_h, 1, stride_align, img_data); } @@ -172,34 +188,33 @@ int vpx_img_set_rect(vpx_image_t *img, unsigned int x, unsigned int y, if (img->fmt & VPX_IMG_FMT_HAS_ALPHA) { img->planes[VPX_PLANE_ALPHA] = data + x * bytes_per_sample + y * img->stride[VPX_PLANE_ALPHA]; - data += img->h * img->stride[VPX_PLANE_ALPHA]; + data += (size_t)img->h * img->stride[VPX_PLANE_ALPHA]; } img->planes[VPX_PLANE_Y] = data + x * bytes_per_sample + y * img->stride[VPX_PLANE_Y]; - data += img->h * img->stride[VPX_PLANE_Y]; + data += (size_t)img->h * img->stride[VPX_PLANE_Y]; + unsigned int uv_x = x >> img->x_chroma_shift; + unsigned int uv_y = y >> img->y_chroma_shift; if (img->fmt == VPX_IMG_FMT_NV12) { img->planes[VPX_PLANE_U] = - data + (x >> img->x_chroma_shift) + - (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_U]; + data + uv_x + uv_y * img->stride[VPX_PLANE_U]; img->planes[VPX_PLANE_V] = img->planes[VPX_PLANE_U] + 1; } else if (!(img->fmt & VPX_IMG_FMT_UV_FLIP)) { img->planes[VPX_PLANE_U] = - data + (x >> img->x_chroma_shift) * bytes_per_sample + - (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_U]; - data += (img->h >> img->y_chroma_shift) * img->stride[VPX_PLANE_U]; + data + uv_x * bytes_per_sample + uv_y * img->stride[VPX_PLANE_U]; + data += + (size_t)(img->h >> img->y_chroma_shift) * img->stride[VPX_PLANE_U]; img->planes[VPX_PLANE_V] = - data + (x >> img->x_chroma_shift) * bytes_per_sample + - (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_V]; + data + uv_x * bytes_per_sample + uv_y * img->stride[VPX_PLANE_V]; } else { img->planes[VPX_PLANE_V] = - data + (x >> img->x_chroma_shift) * bytes_per_sample + - (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_V]; - data += (img->h >> img->y_chroma_shift) * img->stride[VPX_PLANE_V]; + data + uv_x * bytes_per_sample + uv_y * img->stride[VPX_PLANE_V]; + data += + (size_t)(img->h >> img->y_chroma_shift) * img->stride[VPX_PLANE_V]; img->planes[VPX_PLANE_U] = - data + (x >> img->x_chroma_shift) * bytes_per_sample + - (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_U]; + data + uv_x * bytes_per_sample + uv_y * img->stride[VPX_PLANE_U]; } } return 0; diff --git a/media/libvpx/libvpx/vpx/src/vpx_tpl.c b/media/libvpx/libvpx/vpx/src/vpx_tpl.c deleted file mode 100644 index b0687a8135..0000000000 --- a/media/libvpx/libvpx/vpx/src/vpx_tpl.c +++ /dev/null @@ -1,107 +0,0 @@ -/* - * Copyright (c) 2023 The WebM project authors. All Rights Reserved. - * - * Use of this source code is governed by a BSD-style license - * that can be found in the LICENSE file in the root of the source - * tree. An additional intellectual property rights grant can be found - * in the file PATENTS. All contributing project authors may - * be found in the AUTHORS file in the root of the source tree. - */ - -#include <stdlib.h> - -#include "vpx/vpx_codec.h" -#include "vpx/vpx_tpl.h" -#include "vpx_mem/vpx_mem.h" - -#define CHECK_FPRINTF_ERROR(expr) \ - do { \ - if (expr < 0) { \ - return VPX_CODEC_ERROR; \ - } \ - } while (0) - -#define CHECK_FSCANF_ERROR(expr, expected_value) \ - do { \ - if (expr != expected_value) { \ - return VPX_CODEC_ERROR; \ - } \ - } while (0) - -vpx_codec_err_t vpx_write_tpl_gop_stats(FILE *tpl_file, - const VpxTplGopStats *tpl_gop_stats) { - int i; - if (tpl_file == NULL || tpl_gop_stats == NULL) return VPX_CODEC_INVALID_PARAM; - CHECK_FPRINTF_ERROR(fprintf(tpl_file, "%d\n", tpl_gop_stats->size)); - - for (i = 0; i < tpl_gop_stats->size; i++) { - VpxTplFrameStats frame_stats = tpl_gop_stats->frame_stats_list[i]; - const int num_blocks = frame_stats.num_blocks; - int block; - CHECK_FPRINTF_ERROR(fprintf(tpl_file, "%d %d %d\n", frame_stats.frame_width, - frame_stats.frame_height, num_blocks)); - for (block = 0; block < num_blocks; block++) { - VpxTplBlockStats block_stats = frame_stats.block_stats_list[block]; - CHECK_FPRINTF_ERROR( - fprintf(tpl_file, - "%" PRId64 " %" PRId64 " %" PRId16 " %" PRId16 " %" PRId64 - " %" PRId64 " %d\n", - block_stats.inter_cost, block_stats.intra_cost, - block_stats.mv_c, block_stats.mv_r, block_stats.srcrf_dist, - block_stats.srcrf_rate, block_stats.ref_frame_index)); - } - } - - return VPX_CODEC_OK; -} - -vpx_codec_err_t vpx_read_tpl_gop_stats(FILE *tpl_file, - VpxTplGopStats *tpl_gop_stats) { - int i, frame_list_size; - if (tpl_file == NULL || tpl_gop_stats == NULL) return VPX_CODEC_INVALID_PARAM; - CHECK_FSCANF_ERROR(fscanf(tpl_file, "%d\n", &frame_list_size), 1); - tpl_gop_stats->size = frame_list_size; - tpl_gop_stats->frame_stats_list = (VpxTplFrameStats *)vpx_calloc( - frame_list_size, sizeof(tpl_gop_stats->frame_stats_list[0])); - if (tpl_gop_stats->frame_stats_list == NULL) { - return VPX_CODEC_MEM_ERROR; - } - for (i = 0; i < frame_list_size; i++) { - VpxTplFrameStats *frame_stats = &tpl_gop_stats->frame_stats_list[i]; - int num_blocks, width, height, block; - CHECK_FSCANF_ERROR( - fscanf(tpl_file, "%d %d %d\n", &width, &height, &num_blocks), 3); - frame_stats->num_blocks = num_blocks; - frame_stats->frame_width = width; - frame_stats->frame_height = height; - frame_stats->block_stats_list = (VpxTplBlockStats *)vpx_calloc( - num_blocks, sizeof(frame_stats->block_stats_list[0])); - if (frame_stats->block_stats_list == NULL) { - vpx_free_tpl_gop_stats(tpl_gop_stats); - return VPX_CODEC_MEM_ERROR; - } - for (block = 0; block < num_blocks; block++) { - VpxTplBlockStats *block_stats = &frame_stats->block_stats_list[block]; - CHECK_FSCANF_ERROR( - fscanf(tpl_file, - "%" SCNd64 " %" SCNd64 " %" SCNd16 " %" SCNd16 " %" SCNd64 - " %" SCNd64 " %d\n", - &block_stats->inter_cost, &block_stats->intra_cost, - &block_stats->mv_c, &block_stats->mv_r, - &block_stats->srcrf_dist, &block_stats->srcrf_rate, - &block_stats->ref_frame_index), - 7); - } - } - - return VPX_CODEC_OK; -} - -void vpx_free_tpl_gop_stats(VpxTplGopStats *tpl_gop_stats) { - int frame; - if (tpl_gop_stats == NULL) return; - for (frame = 0; frame < tpl_gop_stats->size; frame++) { - vpx_free(tpl_gop_stats->frame_stats_list[frame].block_stats_list); - } - vpx_free(tpl_gop_stats->frame_stats_list); -} diff --git a/media/libvpx/libvpx/vpx/vpx_codec.mk b/media/libvpx/libvpx/vpx/vpx_codec.mk index 25c815ef51..4aec88b300 100644 --- a/media/libvpx/libvpx/vpx/vpx_codec.mk +++ b/media/libvpx/libvpx/vpx/vpx_codec.mk @@ -37,7 +37,6 @@ API_SRCS-yes += internal/vpx_codec_internal.h API_SRCS-yes += internal/vpx_ratectrl_rtc.h API_SRCS-yes += src/vpx_codec.c API_SRCS-yes += src/vpx_image.c -API_SRCS-yes += src/vpx_tpl.c API_SRCS-yes += vpx_codec.h API_SRCS-yes += vpx_codec.mk API_SRCS-yes += vpx_frame_buffer.h diff --git a/media/libvpx/libvpx/vpx/vpx_encoder.h b/media/libvpx/libvpx/vpx/vpx_encoder.h index 809a097d94..740e7aa444 100644 --- a/media/libvpx/libvpx/vpx/vpx_encoder.h +++ b/media/libvpx/libvpx/vpx/vpx_encoder.h @@ -1019,6 +1019,8 @@ typedef unsigned long vpx_enc_deadline_t; * * \param[in] ctx Pointer to this instance's context * \param[in] img Image data to encode, NULL to flush. + * Encoding sample values outside the range + * [0..(1<<img->bit_depth)-1] is undefined behavior. * \param[in] pts Presentation time stamp, in timebase units. * \param[in] duration Duration to show frame, in timebase units. * \param[in] flags Flags to use for encoding this frame. diff --git a/media/libvpx/libvpx/vpx/vpx_ext_ratectrl.h b/media/libvpx/libvpx/vpx/vpx_ext_ratectrl.h index ba12e4f83b..d56e151114 100644 --- a/media/libvpx/libvpx/vpx/vpx_ext_ratectrl.h +++ b/media/libvpx/libvpx/vpx/vpx_ext_ratectrl.h @@ -28,6 +28,15 @@ extern "C" { */ #define VPX_EXT_RATECTRL_ABI_VERSION (5 + VPX_TPL_ABI_VERSION) +/*!\brief This is correspondent to MAX_STATIC_GF_GROUP_LENGTH defined in + * vp9_ratectrl.h + */ +#define VPX_RC_MAX_STATIC_GF_GROUP_LENGTH 250 + +/*!\brief Max number of ref frames returned by the external RC. Correspondent to + * MAX_REF_FRAMES defined in vp9_blockd.h. */ +#define VPX_RC_MAX_REF_FRAMES 4 + /*!\brief The control type of the inference API. * In VPX_RC_QP mode, the external rate control model determines the * quantization parameter (QP) for each frame. @@ -56,6 +65,29 @@ typedef enum vpx_ext_rc_mode { VPX_RC_CQ = 2, } vpx_ext_rc_mode_t; +/*!\brief This is correspondent to FRAME_UPDATE_TYPE defined in vp9_firstpass.h. + */ +typedef enum vpx_rc_frame_update_type { + VPX_RC_INVALID_UPDATE_TYPE = -1, + VPX_RC_KF_UPDATE = 0, + VPX_RC_LF_UPDATE = 1, + VPX_RC_GF_UPDATE = 2, + VPX_RC_ARF_UPDATE = 3, + VPX_RC_OVERLAY_UPDATE = 4, + VPX_RC_MID_OVERLAY_UPDATE = 5, + VPX_RC_USE_BUF_FRAME = 6, +} vpx_rc_frame_update_type_t; + +/*!\brief Name for the ref frames returned by the external RC. Correspondent to + * the ref frames defined in vp9_blockd.h. */ +typedef enum vpx_rc_ref_name { + VPX_RC_INVALID_REF_FRAME = -1, + VPX_RC_INTRA_FRAME = 0, + VPX_RC_LAST_FRAME = 1, + VPX_RC_GOLDEN_FRAME = 2, + VPX_RC_ALTREF_FRAME = 3, +} vpx_rc_ref_name_t; + /*!\brief Abstract rate control model handler * * The encoder will receive the model handler from create_model() defined in @@ -318,75 +350,12 @@ typedef struct vpx_rc_config { int base_qp; /**< base QP for leaf frames, 0-255 */ } vpx_rc_config_t; -/*!\brief Information passed to the external rate control model to - * help make GOP decisions. +/*!\brief Control what ref frame to use and its index. */ -typedef struct vpx_rc_gop_info { - /*! - * Minimum allowed gf interval, fixed for the whole clip. - * Note that it will be modified to match vp9's level constraints - * in the encoder. - * The level constraint is defined in vp9_encoder.c: - * const Vp9LevelSpec vp9_level_defs[VP9_LEVELS]. - */ - int min_gf_interval; - /*! - * Maximum allowed gf interval, fixed for the whole clip. - */ - int max_gf_interval; - /*! - * Minimum allowed gf interval for the current GOP, determined - * by the encoder. - */ - int active_min_gf_interval; - /*! - * Maximum allowed gf interval for the current GOP, determined - * by the encoder. - */ - int active_max_gf_interval; - /*! - * Whether to allow the use of alt ref, determined by the encoder. - * It is fixed for the entire encode. - * See function "is_altref_enabled" in vp9_encoder.h. - */ - int allow_alt_ref; - /*! - * Is the current frame a key frame. - */ - int is_key_frame; - /*! - * Does the previous gop use alt ref or not. - */ - int last_gop_use_alt_ref; - /*! - * Current frame distance to the last keyframe, e.g., if Nth frame is a key, - * then the value of the N+1 th frame is 1. - */ - int frames_since_key; - /*! - * Current frame distance to the next keyframe, e.g. if Nth frame is a key, - * then the value of frame N - 1 is 1. - */ - int frames_to_key; - /*! - * Number of lookahead source frames. - */ - int lag_in_frames; - /*! - * Display index (temporal stamp) of this frame in the whole clip, - * starts from zero. - */ - int show_index; - /*! - * Coding index of this frame in the whole clip, starts from zero. - */ - int coding_index; - /*! - * The index of the current gop, starts from zero, resets to zero - * when a keyframe is set. - */ - int gop_global_index; -} vpx_rc_gop_info_t; +typedef struct vpx_rc_ref_frame { + int index[VPX_RC_MAX_REF_FRAMES]; + vpx_rc_ref_name_t name[VPX_RC_MAX_REF_FRAMES]; +} vpx_rc_ref_frame_t; /*!\brief The decision made by the external rate control model to set the * group of picture. @@ -395,6 +364,14 @@ typedef struct vpx_rc_gop_decision { int gop_coding_frames; /**< The number of frames of this GOP */ int use_alt_ref; /**< Whether to use alt ref for this GOP */ int use_key_frame; /**< Whether to set key frame for this GOP */ + // Frame type for each frame in this GOP. + // This will be populated to |update_type| in GF_GROUP defined in + // vp9_firstpass.h + vpx_rc_frame_update_type_t update_type[VPX_RC_MAX_STATIC_GF_GROUP_LENGTH + 2]; + // Ref frame buffer index to be updated for each frame in this GOP. + int update_ref_index[VPX_RC_MAX_STATIC_GF_GROUP_LENGTH + 2]; + // Ref frame list to be used for each frame in this GOP. + vpx_rc_ref_frame_t ref_frame_list[VPX_RC_MAX_STATIC_GF_GROUP_LENGTH + 2]; } vpx_rc_gop_decision_t; /*!\brief Create an external rate control model callback prototype diff --git a/media/libvpx/libvpx/vpx/vpx_image.h b/media/libvpx/libvpx/vpx/vpx_image.h index 1adc9b9d9e..af8bf8e6c3 100644 --- a/media/libvpx/libvpx/vpx/vpx_image.h +++ b/media/libvpx/libvpx/vpx/vpx_image.h @@ -64,8 +64,12 @@ typedef enum vpx_color_space { /*!\brief List of supported color range */ typedef enum vpx_color_range { - VPX_CR_STUDIO_RANGE = 0, /**< Y [16..235], UV [16..240] */ - VPX_CR_FULL_RANGE = 1 /**< YUV/RGB [0..255] */ + VPX_CR_STUDIO_RANGE = 0, /**<- Y [16..235], UV [16..240] (bit depth 8) */ + /**<- Y [64..940], UV [64..960] (bit depth 10) */ + /**<- Y [256..3760], UV [256..3840] (bit depth 12) */ + VPX_CR_FULL_RANGE = 1 /**<- YUV/RGB [0..255] (bit depth 8) */ + /**<- YUV/RGB [0..1023] (bit depth 10) */ + /**<- YUV/RGB [0..4095] (bit depth 12) */ } vpx_color_range_t; /**< alias for enum vpx_color_range */ /**\brief Image Descriptor */ @@ -132,10 +136,13 @@ typedef struct vpx_image_rect { * is NULL, the storage for the descriptor will be * allocated on the heap. * \param[in] fmt Format for the image - * \param[in] d_w Width of the image - * \param[in] d_h Height of the image + * \param[in] d_w Width of the image. Must not exceed 0x08000000 + * (2^27). + * \param[in] d_h Height of the image. Must not exceed 0x08000000 + * (2^27). * \param[in] align Alignment, in bytes, of the image buffer and - * each row in the image(stride). + * each row in the image (stride). Must not exceed + * 65536. * * \return Returns a pointer to the initialized image descriptor. If the img * parameter is non-null, the value of the img parameter will be @@ -155,9 +162,12 @@ vpx_image_t *vpx_img_alloc(vpx_image_t *img, vpx_img_fmt_t fmt, * parameter is NULL, the storage for the descriptor * will be allocated on the heap. * \param[in] fmt Format for the image - * \param[in] d_w Width of the image - * \param[in] d_h Height of the image - * \param[in] stride_align Alignment, in bytes, of each row in the image. + * \param[in] d_w Width of the image. Must not exceed 0x08000000 + * (2^27). + * \param[in] d_h Height of the image. Must not exceed 0x08000000 + * (2^27). + * \param[in] stride_align Alignment, in bytes, of each row in the image + * (stride). Must not exceed 65536. * \param[in] img_data Storage to use for the image * * \return Returns a pointer to the initialized image descriptor. If the img diff --git a/media/libvpx/libvpx/vpx/vpx_tpl.h b/media/libvpx/libvpx/vpx/vpx_tpl.h index 7e4c9ab7e1..e14eefcdce 100644 --- a/media/libvpx/libvpx/vpx/vpx_tpl.h +++ b/media/libvpx/libvpx/vpx/vpx_tpl.h @@ -15,8 +15,6 @@ #ifndef VPX_VPX_VPX_TPL_H_ #define VPX_VPX_VPX_TPL_H_ -#include <stdio.h> - #include "./vpx_integer.h" #include "./vpx_codec.h" @@ -32,7 +30,7 @@ extern "C" { * types, removing or reassigning enums, adding/removing/rearranging * fields to structures */ -#define VPX_TPL_ABI_VERSION (3) /**<\hideinitializer*/ +#define VPX_TPL_ABI_VERSION 4 /**<\hideinitializer*/ /*!\brief Temporal dependency model stats for each block before propagation */ typedef struct VpxTplBlockStats { @@ -63,40 +61,6 @@ typedef struct VpxTplGopStats { VpxTplFrameStats *frame_stats_list; /**< List of tpl stats for each frame */ } VpxTplGopStats; -/*!\brief Write VpxTplGopStats to file - * - * Accepts an opened file handle and writes \p tpl_gop_stats. - * - * \param[in] tpl_file A FILE pointer that's already been opened. - * \param[in] tpl_gop_stats VpxTplGopStats that contains TPL stats for the - * whole GOP. - * - * \return VPX_CODEC_OK if TPL stats are successfully written. - */ -vpx_codec_err_t vpx_write_tpl_gop_stats(FILE *tpl_file, - const VpxTplGopStats *tpl_gop_stats); - -/*!\brief Read VpxTplGopStats from file - * - * Accepts an opened file handle and reads TPL stats and stores them into - * \p tpl_gop_stats. Allocates memory for TPL stats. - * - * \param[in] tpl_file A FILE pointer that's already been opened. - * \param[out] tpl_gop_stats VpxTplGopStats that contains TPL stats for the - * whole GOP. - * - * \return VPX_CODEC_OK if TPL stats are successfully read from file. - */ -vpx_codec_err_t vpx_read_tpl_gop_stats(FILE *tpl_file, - VpxTplGopStats *tpl_gop_stats); - -/*!\brief Free the memory allocated for VpxTplGopStats - * - * \param[in] tpl_gop_stats VpxTplGopStats that contains TPL stats for the - * whole GOP. - */ -void vpx_free_tpl_gop_stats(VpxTplGopStats *tpl_gop_stats); - #ifdef __cplusplus } // extern "C" #endif diff --git a/media/libvpx/libvpx/vpx_dsp/arm/highbd_convolve8_neon.h b/media/libvpx/libvpx/vpx_dsp/arm/highbd_convolve8_neon.h new file mode 100644 index 0000000000..ccc4a797b7 --- /dev/null +++ b/media/libvpx/libvpx/vpx_dsp/arm/highbd_convolve8_neon.h @@ -0,0 +1,46 @@ +/* + * Copyright (c) 2024 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#ifndef VPX_VPX_DSP_ARM_HIGHBD_CONVOLVE8_NEON_H_ +#define VPX_VPX_DSP_ARM_HIGHBD_CONVOLVE8_NEON_H_ + +#include <arm_neon.h> + +static INLINE uint16x4_t highbd_convolve4_4_neon( + const int16x4_t s0, const int16x4_t s1, const int16x4_t s2, + const int16x4_t s3, const int16x4_t filters, const uint16x4_t max) { + int32x4_t sum = vmull_lane_s16(s0, filters, 0); + sum = vmlal_lane_s16(sum, s1, filters, 1); + sum = vmlal_lane_s16(sum, s2, filters, 2); + sum = vmlal_lane_s16(sum, s3, filters, 3); + + uint16x4_t res = vqrshrun_n_s32(sum, FILTER_BITS); + return vmin_u16(res, max); +} + +static INLINE uint16x8_t highbd_convolve4_8_neon( + const int16x8_t s0, const int16x8_t s1, const int16x8_t s2, + const int16x8_t s3, const int16x4_t filters, const uint16x8_t max) { + int32x4_t sum0 = vmull_lane_s16(vget_low_s16(s0), filters, 0); + sum0 = vmlal_lane_s16(sum0, vget_low_s16(s1), filters, 1); + sum0 = vmlal_lane_s16(sum0, vget_low_s16(s2), filters, 2); + sum0 = vmlal_lane_s16(sum0, vget_low_s16(s3), filters, 3); + + int32x4_t sum1 = vmull_lane_s16(vget_high_s16(s0), filters, 0); + sum1 = vmlal_lane_s16(sum1, vget_high_s16(s1), filters, 1); + sum1 = vmlal_lane_s16(sum1, vget_high_s16(s2), filters, 2); + sum1 = vmlal_lane_s16(sum1, vget_high_s16(s3), filters, 3); + + uint16x8_t res = vcombine_u16(vqrshrun_n_s32(sum0, FILTER_BITS), + vqrshrun_n_s32(sum1, FILTER_BITS)); + return vminq_u16(res, max); +} + +#endif // VPX_VPX_DSP_ARM_HIGHBD_CONVOLVE8_NEON_H_ diff --git a/media/libvpx/libvpx/vpx_dsp/arm/highbd_convolve8_sve.h b/media/libvpx/libvpx/vpx_dsp/arm/highbd_convolve8_sve.h new file mode 100644 index 0000000000..bc90d9b4dd --- /dev/null +++ b/media/libvpx/libvpx/vpx_dsp/arm/highbd_convolve8_sve.h @@ -0,0 +1,99 @@ +/* + * Copyright (c) 2024 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#ifndef VPX_VPX_DSP_ARM_HIGHBD_CONVOLVE8_SVE_H_ +#define VPX_VPX_DSP_ARM_HIGHBD_CONVOLVE8_SVE_H_ + +#include <arm_neon.h> + +#include "vpx_dsp/arm/vpx_neon_sve_bridge.h" + +static INLINE uint16x4_t highbd_convolve4_4_sve(const int16x4_t s[4], + const int16x8_t filter, + const uint16x4_t max) { + int16x8_t s01 = vcombine_s16(s[0], s[1]); + int16x8_t s23 = vcombine_s16(s[2], s[3]); + + int64x2_t sum01 = vpx_dotq_lane_s16(vdupq_n_s64(0), s01, filter, 0); + int64x2_t sum23 = vpx_dotq_lane_s16(vdupq_n_s64(0), s23, filter, 0); + + int32x4_t res_s32 = vcombine_s32(vmovn_s64(sum01), vmovn_s64(sum23)); + + uint16x4_t res_u16 = vqrshrun_n_s32(res_s32, FILTER_BITS); + return vmin_u16(res_u16, max); +} + +static INLINE uint16x8_t highbd_convolve4_8_sve(const int16x8_t s[4], + const int16x8_t filter, + const uint16x8_t max, + uint16x8_t idx) { + int64x2_t sum04 = vpx_dotq_lane_s16(vdupq_n_s64(0), s[0], filter, 0); + int64x2_t sum15 = vpx_dotq_lane_s16(vdupq_n_s64(0), s[1], filter, 0); + int64x2_t sum26 = vpx_dotq_lane_s16(vdupq_n_s64(0), s[2], filter, 0); + int64x2_t sum37 = vpx_dotq_lane_s16(vdupq_n_s64(0), s[3], filter, 0); + + int32x4_t res0 = vcombine_s32(vmovn_s64(sum04), vmovn_s64(sum15)); + int32x4_t res1 = vcombine_s32(vmovn_s64(sum26), vmovn_s64(sum37)); + + uint16x8_t res = vcombine_u16(vqrshrun_n_s32(res0, FILTER_BITS), + vqrshrun_n_s32(res1, FILTER_BITS)); + + res = vpx_tbl_u16(res, idx); + + return vminq_u16(res, max); +} + +static INLINE uint16x4_t highbd_convolve8_4(const int16x8_t s[4], + const int16x8_t filter, + const uint16x4_t max) { + int64x2_t sum[4]; + + sum[0] = vpx_dotq_s16(vdupq_n_s64(0), s[0], filter); + sum[1] = vpx_dotq_s16(vdupq_n_s64(0), s[1], filter); + sum[2] = vpx_dotq_s16(vdupq_n_s64(0), s[2], filter); + sum[3] = vpx_dotq_s16(vdupq_n_s64(0), s[3], filter); + + sum[0] = vpaddq_s64(sum[0], sum[1]); + sum[2] = vpaddq_s64(sum[2], sum[3]); + + int32x4_t res_s32 = vcombine_s32(vmovn_s64(sum[0]), vmovn_s64(sum[2])); + + uint16x4_t res_u16 = vqrshrun_n_s32(res_s32, FILTER_BITS); + return vmin_u16(res_u16, max); +} + +static INLINE uint16x8_t highbd_convolve8_8(const int16x8_t s[8], + const int16x8_t filter, + const uint16x8_t max) { + int64x2_t sum[8]; + + sum[0] = vpx_dotq_s16(vdupq_n_s64(0), s[0], filter); + sum[1] = vpx_dotq_s16(vdupq_n_s64(0), s[1], filter); + sum[2] = vpx_dotq_s16(vdupq_n_s64(0), s[2], filter); + sum[3] = vpx_dotq_s16(vdupq_n_s64(0), s[3], filter); + sum[4] = vpx_dotq_s16(vdupq_n_s64(0), s[4], filter); + sum[5] = vpx_dotq_s16(vdupq_n_s64(0), s[5], filter); + sum[6] = vpx_dotq_s16(vdupq_n_s64(0), s[6], filter); + sum[7] = vpx_dotq_s16(vdupq_n_s64(0), s[7], filter); + + int64x2_t sum01 = vpaddq_s64(sum[0], sum[1]); + int64x2_t sum23 = vpaddq_s64(sum[2], sum[3]); + int64x2_t sum45 = vpaddq_s64(sum[4], sum[5]); + int64x2_t sum67 = vpaddq_s64(sum[6], sum[7]); + + int32x4_t res0 = vcombine_s32(vmovn_s64(sum01), vmovn_s64(sum23)); + int32x4_t res1 = vcombine_s32(vmovn_s64(sum45), vmovn_s64(sum67)); + + uint16x8_t res = vcombine_u16(vqrshrun_n_s32(res0, FILTER_BITS), + vqrshrun_n_s32(res1, FILTER_BITS)); + return vminq_u16(res, max); +} + +#endif // VPX_VPX_DSP_ARM_HIGHBD_CONVOLVE8_SVE_H_ diff --git a/media/libvpx/libvpx/vpx_dsp/arm/highbd_vpx_convolve8_neon.c b/media/libvpx/libvpx/vpx_dsp/arm/highbd_vpx_convolve8_neon.c index b5a944d299..cc6307f923 100644 --- a/media/libvpx/libvpx/vpx_dsp/arm/highbd_vpx_convolve8_neon.c +++ b/media/libvpx/libvpx/vpx_dsp/arm/highbd_vpx_convolve8_neon.c @@ -14,42 +14,13 @@ #include "./vpx_config.h" #include "./vpx_dsp_rtcd.h" #include "vpx/vpx_integer.h" +#include "vpx_dsp/arm/highbd_convolve8_neon.h" #include "vpx_dsp/arm/mem_neon.h" #include "vpx_dsp/arm/transpose_neon.h" #include "vpx_dsp/vpx_dsp_common.h" #include "vpx_dsp/vpx_filter.h" #include "vpx_ports/mem.h" -static INLINE uint16x4_t highbd_convolve4_4( - const int16x4_t s0, const int16x4_t s1, const int16x4_t s2, - const int16x4_t s3, const int16x4_t filters, const uint16x4_t max) { - int32x4_t sum = vmull_lane_s16(s0, filters, 0); - sum = vmlal_lane_s16(sum, s1, filters, 1); - sum = vmlal_lane_s16(sum, s2, filters, 2); - sum = vmlal_lane_s16(sum, s3, filters, 3); - - uint16x4_t res = vqrshrun_n_s32(sum, FILTER_BITS); - return vmin_u16(res, max); -} - -static INLINE uint16x8_t highbd_convolve4_8( - const int16x8_t s0, const int16x8_t s1, const int16x8_t s2, - const int16x8_t s3, const int16x4_t filters, const uint16x8_t max) { - int32x4_t sum0 = vmull_lane_s16(vget_low_s16(s0), filters, 0); - sum0 = vmlal_lane_s16(sum0, vget_low_s16(s1), filters, 1); - sum0 = vmlal_lane_s16(sum0, vget_low_s16(s2), filters, 2); - sum0 = vmlal_lane_s16(sum0, vget_low_s16(s3), filters, 3); - - int32x4_t sum1 = vmull_lane_s16(vget_high_s16(s0), filters, 0); - sum1 = vmlal_lane_s16(sum1, vget_high_s16(s1), filters, 1); - sum1 = vmlal_lane_s16(sum1, vget_high_s16(s2), filters, 2); - sum1 = vmlal_lane_s16(sum1, vget_high_s16(s3), filters, 3); - - uint16x8_t res = vcombine_u16(vqrshrun_n_s32(sum0, FILTER_BITS), - vqrshrun_n_s32(sum1, FILTER_BITS)); - return vminq_u16(res, max); -} - static INLINE uint16x4_t highbd_convolve8_4(const int16x4_t s0, const int16x4_t s1, const int16x4_t s2, const int16x4_t s3, const int16x4_t s4, const int16x4_t s5, @@ -118,13 +89,13 @@ static INLINE void highbd_convolve_4tap_horiz_neon( load_s16_4x4(s + 3 * src_stride, 1, &s3[0], &s3[1], &s3[2], &s3[3]); uint16x4_t d0 = - highbd_convolve4_4(s0[0], s0[1], s0[2], s0[3], filter, max); + highbd_convolve4_4_neon(s0[0], s0[1], s0[2], s0[3], filter, max); uint16x4_t d1 = - highbd_convolve4_4(s1[0], s1[1], s1[2], s1[3], filter, max); + highbd_convolve4_4_neon(s1[0], s1[1], s1[2], s1[3], filter, max); uint16x4_t d2 = - highbd_convolve4_4(s2[0], s2[1], s2[2], s2[3], filter, max); + highbd_convolve4_4_neon(s2[0], s2[1], s2[2], s2[3], filter, max); uint16x4_t d3 = - highbd_convolve4_4(s3[0], s3[1], s3[2], s3[3], filter, max); + highbd_convolve4_4_neon(s3[0], s3[1], s3[2], s3[3], filter, max); store_u16_4x4(d, dst_stride, d0, d1, d2, d3); @@ -148,13 +119,13 @@ static INLINE void highbd_convolve_4tap_horiz_neon( load_s16_8x4(s + 3 * src_stride, 1, &s3[0], &s3[1], &s3[2], &s3[3]); uint16x8_t d0 = - highbd_convolve4_8(s0[0], s0[1], s0[2], s0[3], filter, max); + highbd_convolve4_8_neon(s0[0], s0[1], s0[2], s0[3], filter, max); uint16x8_t d1 = - highbd_convolve4_8(s1[0], s1[1], s1[2], s1[3], filter, max); + highbd_convolve4_8_neon(s1[0], s1[1], s1[2], s1[3], filter, max); uint16x8_t d2 = - highbd_convolve4_8(s2[0], s2[1], s2[2], s2[3], filter, max); + highbd_convolve4_8_neon(s2[0], s2[1], s2[2], s2[3], filter, max); uint16x8_t d3 = - highbd_convolve4_8(s3[0], s3[1], s3[2], s3[3], filter, max); + highbd_convolve4_8_neon(s3[0], s3[1], s3[2], s3[3], filter, max); store_u16_8x4(d, dst_stride, d0, d1, d2, d3); @@ -393,10 +364,10 @@ static INLINE void highbd_convolve_4tap_vert_neon( int16x4_t s3, s4, s5, s6; load_s16_4x4(s, src_stride, &s3, &s4, &s5, &s6); - uint16x4_t d0 = highbd_convolve4_4(s0, s1, s2, s3, filter, max); - uint16x4_t d1 = highbd_convolve4_4(s1, s2, s3, s4, filter, max); - uint16x4_t d2 = highbd_convolve4_4(s2, s3, s4, s5, filter, max); - uint16x4_t d3 = highbd_convolve4_4(s3, s4, s5, s6, filter, max); + uint16x4_t d0 = highbd_convolve4_4_neon(s0, s1, s2, s3, filter, max); + uint16x4_t d1 = highbd_convolve4_4_neon(s1, s2, s3, s4, filter, max); + uint16x4_t d2 = highbd_convolve4_4_neon(s2, s3, s4, s5, filter, max); + uint16x4_t d3 = highbd_convolve4_4_neon(s3, s4, s5, s6, filter, max); store_u16_4x4(d, dst_stride, d0, d1, d2, d3); @@ -424,10 +395,10 @@ static INLINE void highbd_convolve_4tap_vert_neon( int16x8_t s3, s4, s5, s6; load_s16_8x4(s, src_stride, &s3, &s4, &s5, &s6); - uint16x8_t d0 = highbd_convolve4_8(s0, s1, s2, s3, filter, max); - uint16x8_t d1 = highbd_convolve4_8(s1, s2, s3, s4, filter, max); - uint16x8_t d2 = highbd_convolve4_8(s2, s3, s4, s5, filter, max); - uint16x8_t d3 = highbd_convolve4_8(s3, s4, s5, s6, filter, max); + uint16x8_t d0 = highbd_convolve4_8_neon(s0, s1, s2, s3, filter, max); + uint16x8_t d1 = highbd_convolve4_8_neon(s1, s2, s3, s4, filter, max); + uint16x8_t d2 = highbd_convolve4_8_neon(s2, s3, s4, s5, filter, max); + uint16x8_t d3 = highbd_convolve4_8_neon(s3, s4, s5, s6, filter, max); store_u16_8x4(d, dst_stride, d0, d1, d2, d3); @@ -686,12 +657,12 @@ static INLINE void highbd_convolve_2d_4tap_neon( load_s16_4x4(s + 1 * src_stride, 1, &h_s1[0], &h_s1[1], &h_s1[2], &h_s1[3]); load_s16_4x4(s + 2 * src_stride, 1, &h_s2[0], &h_s2[1], &h_s2[2], &h_s2[3]); - int16x4_t v_s0 = vreinterpret_s16_u16( - highbd_convolve4_4(h_s0[0], h_s0[1], h_s0[2], h_s0[3], x_filter, max)); - int16x4_t v_s1 = vreinterpret_s16_u16( - highbd_convolve4_4(h_s1[0], h_s1[1], h_s1[2], h_s1[3], x_filter, max)); - int16x4_t v_s2 = vreinterpret_s16_u16( - highbd_convolve4_4(h_s2[0], h_s2[1], h_s2[2], h_s2[3], x_filter, max)); + int16x4_t v_s0 = vreinterpret_s16_u16(highbd_convolve4_4_neon( + h_s0[0], h_s0[1], h_s0[2], h_s0[3], x_filter, max)); + int16x4_t v_s1 = vreinterpret_s16_u16(highbd_convolve4_4_neon( + h_s1[0], h_s1[1], h_s1[2], h_s1[3], x_filter, max)); + int16x4_t v_s2 = vreinterpret_s16_u16(highbd_convolve4_4_neon( + h_s2[0], h_s2[1], h_s2[2], h_s2[3], x_filter, max)); s += 3 * src_stride; @@ -706,19 +677,23 @@ static INLINE void highbd_convolve_2d_4tap_neon( load_s16_4x4(s + 3 * src_stride, 1, &h_s6[0], &h_s6[1], &h_s6[2], &h_s6[3]); - int16x4_t v_s3 = vreinterpret_s16_u16(highbd_convolve4_4( + int16x4_t v_s3 = vreinterpret_s16_u16(highbd_convolve4_4_neon( h_s3[0], h_s3[1], h_s3[2], h_s3[3], x_filter, max)); - int16x4_t v_s4 = vreinterpret_s16_u16(highbd_convolve4_4( + int16x4_t v_s4 = vreinterpret_s16_u16(highbd_convolve4_4_neon( h_s4[0], h_s4[1], h_s4[2], h_s4[3], x_filter, max)); - int16x4_t v_s5 = vreinterpret_s16_u16(highbd_convolve4_4( + int16x4_t v_s5 = vreinterpret_s16_u16(highbd_convolve4_4_neon( h_s5[0], h_s5[1], h_s5[2], h_s5[3], x_filter, max)); - int16x4_t v_s6 = vreinterpret_s16_u16(highbd_convolve4_4( + int16x4_t v_s6 = vreinterpret_s16_u16(highbd_convolve4_4_neon( h_s6[0], h_s6[1], h_s6[2], h_s6[3], x_filter, max)); - uint16x4_t d0 = highbd_convolve4_4(v_s0, v_s1, v_s2, v_s3, y_filter, max); - uint16x4_t d1 = highbd_convolve4_4(v_s1, v_s2, v_s3, v_s4, y_filter, max); - uint16x4_t d2 = highbd_convolve4_4(v_s2, v_s3, v_s4, v_s5, y_filter, max); - uint16x4_t d3 = highbd_convolve4_4(v_s3, v_s4, v_s5, v_s6, y_filter, max); + uint16x4_t d0 = + highbd_convolve4_4_neon(v_s0, v_s1, v_s2, v_s3, y_filter, max); + uint16x4_t d1 = + highbd_convolve4_4_neon(v_s1, v_s2, v_s3, v_s4, y_filter, max); + uint16x4_t d2 = + highbd_convolve4_4_neon(v_s2, v_s3, v_s4, v_s5, y_filter, max); + uint16x4_t d3 = + highbd_convolve4_4_neon(v_s3, v_s4, v_s5, v_s6, y_filter, max); store_u16_4x4(d, dst_stride, d0, d1, d2, d3); @@ -745,12 +720,12 @@ static INLINE void highbd_convolve_2d_4tap_neon( load_s16_8x4(s + 1 * src_stride, 1, &h_s1[0], &h_s1[1], &h_s1[2], &h_s1[3]); load_s16_8x4(s + 2 * src_stride, 1, &h_s2[0], &h_s2[1], &h_s2[2], &h_s2[3]); - int16x8_t v_s0 = vreinterpretq_s16_u16( - highbd_convolve4_8(h_s0[0], h_s0[1], h_s0[2], h_s0[3], x_filter, max)); - int16x8_t v_s1 = vreinterpretq_s16_u16( - highbd_convolve4_8(h_s1[0], h_s1[1], h_s1[2], h_s1[3], x_filter, max)); - int16x8_t v_s2 = vreinterpretq_s16_u16( - highbd_convolve4_8(h_s2[0], h_s2[1], h_s2[2], h_s2[3], x_filter, max)); + int16x8_t v_s0 = vreinterpretq_s16_u16(highbd_convolve4_8_neon( + h_s0[0], h_s0[1], h_s0[2], h_s0[3], x_filter, max)); + int16x8_t v_s1 = vreinterpretq_s16_u16(highbd_convolve4_8_neon( + h_s1[0], h_s1[1], h_s1[2], h_s1[3], x_filter, max)); + int16x8_t v_s2 = vreinterpretq_s16_u16(highbd_convolve4_8_neon( + h_s2[0], h_s2[1], h_s2[2], h_s2[3], x_filter, max)); s += 3 * src_stride; @@ -765,19 +740,23 @@ static INLINE void highbd_convolve_2d_4tap_neon( load_s16_8x4(s + 3 * src_stride, 1, &h_s6[0], &h_s6[1], &h_s6[2], &h_s6[3]); - int16x8_t v_s3 = vreinterpretq_s16_u16(highbd_convolve4_8( + int16x8_t v_s3 = vreinterpretq_s16_u16(highbd_convolve4_8_neon( h_s3[0], h_s3[1], h_s3[2], h_s3[3], x_filter, max)); - int16x8_t v_s4 = vreinterpretq_s16_u16(highbd_convolve4_8( + int16x8_t v_s4 = vreinterpretq_s16_u16(highbd_convolve4_8_neon( h_s4[0], h_s4[1], h_s4[2], h_s4[3], x_filter, max)); - int16x8_t v_s5 = vreinterpretq_s16_u16(highbd_convolve4_8( + int16x8_t v_s5 = vreinterpretq_s16_u16(highbd_convolve4_8_neon( h_s5[0], h_s5[1], h_s5[2], h_s5[3], x_filter, max)); - int16x8_t v_s6 = vreinterpretq_s16_u16(highbd_convolve4_8( + int16x8_t v_s6 = vreinterpretq_s16_u16(highbd_convolve4_8_neon( h_s6[0], h_s6[1], h_s6[2], h_s6[3], x_filter, max)); - uint16x8_t d0 = highbd_convolve4_8(v_s0, v_s1, v_s2, v_s3, y_filter, max); - uint16x8_t d1 = highbd_convolve4_8(v_s1, v_s2, v_s3, v_s4, y_filter, max); - uint16x8_t d2 = highbd_convolve4_8(v_s2, v_s3, v_s4, v_s5, y_filter, max); - uint16x8_t d3 = highbd_convolve4_8(v_s3, v_s4, v_s5, v_s6, y_filter, max); + uint16x8_t d0 = + highbd_convolve4_8_neon(v_s0, v_s1, v_s2, v_s3, y_filter, max); + uint16x8_t d1 = + highbd_convolve4_8_neon(v_s1, v_s2, v_s3, v_s4, y_filter, max); + uint16x8_t d2 = + highbd_convolve4_8_neon(v_s2, v_s3, v_s4, v_s5, y_filter, max); + uint16x8_t d3 = + highbd_convolve4_8_neon(v_s3, v_s4, v_s5, v_s6, y_filter, max); store_u16_8x4(d, dst_stride, d0, d1, d2, d3); diff --git a/media/libvpx/libvpx/vpx_dsp/arm/highbd_vpx_convolve8_sve.c b/media/libvpx/libvpx/vpx_dsp/arm/highbd_vpx_convolve8_sve.c index 7fc0a57c90..f909e06a18 100644 --- a/media/libvpx/libvpx/vpx_dsp/arm/highbd_vpx_convolve8_sve.c +++ b/media/libvpx/libvpx/vpx_dsp/arm/highbd_vpx_convolve8_sve.c @@ -15,6 +15,7 @@ #include "./vpx_dsp_rtcd.h" #include "vpx/vpx_integer.h" +#include "vpx_dsp/arm/highbd_convolve8_sve.h" #include "vpx_dsp/arm/mem_neon.h" #include "vpx_dsp/arm/transpose_neon.h" #include "vpx_dsp/arm/vpx_neon_sve_bridge.h" @@ -22,87 +23,6 @@ DECLARE_ALIGNED(16, static const uint16_t, kTblConv4_8[8]) = { 0, 2, 4, 6, 1, 3, 5, 7 }; -static INLINE uint16x4_t highbd_convolve4_4(const int16x4_t s[4], - const int16x8_t filter, - const uint16x4_t max) { - int16x8_t s01 = vcombine_s16(s[0], s[1]); - int16x8_t s23 = vcombine_s16(s[2], s[3]); - - int64x2_t sum01 = vpx_dotq_lane_s16(vdupq_n_s64(0), s01, filter, 0); - int64x2_t sum23 = vpx_dotq_lane_s16(vdupq_n_s64(0), s23, filter, 0); - - int32x4_t res_s32 = vcombine_s32(vmovn_s64(sum01), vmovn_s64(sum23)); - - uint16x4_t res_u16 = vqrshrun_n_s32(res_s32, FILTER_BITS); - return vmin_u16(res_u16, max); -} - -static INLINE uint16x8_t highbd_convolve4_8(const int16x8_t s[4], - const int16x8_t filter, - const uint16x8_t max, - uint16x8_t idx) { - int64x2_t sum04 = vpx_dotq_lane_s16(vdupq_n_s64(0), s[0], filter, 0); - int64x2_t sum15 = vpx_dotq_lane_s16(vdupq_n_s64(0), s[1], filter, 0); - int64x2_t sum26 = vpx_dotq_lane_s16(vdupq_n_s64(0), s[2], filter, 0); - int64x2_t sum37 = vpx_dotq_lane_s16(vdupq_n_s64(0), s[3], filter, 0); - - int32x4_t res0 = vcombine_s32(vmovn_s64(sum04), vmovn_s64(sum15)); - int32x4_t res1 = vcombine_s32(vmovn_s64(sum26), vmovn_s64(sum37)); - - uint16x8_t res = vcombine_u16(vqrshrun_n_s32(res0, FILTER_BITS), - vqrshrun_n_s32(res1, FILTER_BITS)); - - res = vpx_tbl_u16(res, idx); - - return vminq_u16(res, max); -} - -static INLINE uint16x4_t highbd_convolve8_4(const int16x8_t s[4], - const int16x8_t filter, - const uint16x4_t max) { - int64x2_t sum[4]; - - sum[0] = vpx_dotq_s16(vdupq_n_s64(0), s[0], filter); - sum[1] = vpx_dotq_s16(vdupq_n_s64(0), s[1], filter); - sum[2] = vpx_dotq_s16(vdupq_n_s64(0), s[2], filter); - sum[3] = vpx_dotq_s16(vdupq_n_s64(0), s[3], filter); - - sum[0] = vpaddq_s64(sum[0], sum[1]); - sum[2] = vpaddq_s64(sum[2], sum[3]); - - int32x4_t res_s32 = vcombine_s32(vmovn_s64(sum[0]), vmovn_s64(sum[2])); - - uint16x4_t res_u16 = vqrshrun_n_s32(res_s32, FILTER_BITS); - return vmin_u16(res_u16, max); -} - -static INLINE uint16x8_t highbd_convolve8_8(const int16x8_t s[8], - const int16x8_t filter, - const uint16x8_t max) { - int64x2_t sum[8]; - - sum[0] = vpx_dotq_s16(vdupq_n_s64(0), s[0], filter); - sum[1] = vpx_dotq_s16(vdupq_n_s64(0), s[1], filter); - sum[2] = vpx_dotq_s16(vdupq_n_s64(0), s[2], filter); - sum[3] = vpx_dotq_s16(vdupq_n_s64(0), s[3], filter); - sum[4] = vpx_dotq_s16(vdupq_n_s64(0), s[4], filter); - sum[5] = vpx_dotq_s16(vdupq_n_s64(0), s[5], filter); - sum[6] = vpx_dotq_s16(vdupq_n_s64(0), s[6], filter); - sum[7] = vpx_dotq_s16(vdupq_n_s64(0), s[7], filter); - - int64x2_t sum01 = vpaddq_s64(sum[0], sum[1]); - int64x2_t sum23 = vpaddq_s64(sum[2], sum[3]); - int64x2_t sum45 = vpaddq_s64(sum[4], sum[5]); - int64x2_t sum67 = vpaddq_s64(sum[6], sum[7]); - - int32x4_t res0 = vcombine_s32(vmovn_s64(sum01), vmovn_s64(sum23)); - int32x4_t res1 = vcombine_s32(vmovn_s64(sum45), vmovn_s64(sum67)); - - uint16x8_t res = vcombine_u16(vqrshrun_n_s32(res0, FILTER_BITS), - vqrshrun_n_s32(res1, FILTER_BITS)); - return vminq_u16(res, max); -} - static INLINE void highbd_convolve_4tap_horiz_sve( const uint16_t *src, ptrdiff_t src_stride, uint16_t *dst, ptrdiff_t dst_stride, int w, int h, const int16x4_t filters, int bd) { @@ -120,10 +40,10 @@ static INLINE void highbd_convolve_4tap_horiz_sve( load_s16_4x4(s + 2 * src_stride, 1, &s2[0], &s2[1], &s2[2], &s2[3]); load_s16_4x4(s + 3 * src_stride, 1, &s3[0], &s3[1], &s3[2], &s3[3]); - uint16x4_t d0 = highbd_convolve4_4(s0, filter, max); - uint16x4_t d1 = highbd_convolve4_4(s1, filter, max); - uint16x4_t d2 = highbd_convolve4_4(s2, filter, max); - uint16x4_t d3 = highbd_convolve4_4(s3, filter, max); + uint16x4_t d0 = highbd_convolve4_4_sve(s0, filter, max); + uint16x4_t d1 = highbd_convolve4_4_sve(s1, filter, max); + uint16x4_t d2 = highbd_convolve4_4_sve(s2, filter, max); + uint16x4_t d3 = highbd_convolve4_4_sve(s3, filter, max); store_u16_4x4(d, dst_stride, d0, d1, d2, d3); @@ -147,10 +67,10 @@ static INLINE void highbd_convolve_4tap_horiz_sve( load_s16_8x4(s + 2 * src_stride, 1, &s2[0], &s2[1], &s2[2], &s2[3]); load_s16_8x4(s + 3 * src_stride, 1, &s3[0], &s3[1], &s3[2], &s3[3]); - uint16x8_t d0 = highbd_convolve4_8(s0, filter, max, idx); - uint16x8_t d1 = highbd_convolve4_8(s1, filter, max, idx); - uint16x8_t d2 = highbd_convolve4_8(s2, filter, max, idx); - uint16x8_t d3 = highbd_convolve4_8(s3, filter, max, idx); + uint16x8_t d0 = highbd_convolve4_8_sve(s0, filter, max, idx); + uint16x8_t d1 = highbd_convolve4_8_sve(s1, filter, max, idx); + uint16x8_t d2 = highbd_convolve4_8_sve(s2, filter, max, idx); + uint16x8_t d3 = highbd_convolve4_8_sve(s3, filter, max, idx); store_u16_8x4(d, dst_stride, d0, d1, d2, d3); diff --git a/media/libvpx/libvpx/vpx_dsp/arm/highbd_vpx_convolve8_sve2.c b/media/libvpx/libvpx/vpx_dsp/arm/highbd_vpx_convolve8_sve2.c index 4ed7718f7d..6f0581d3ed 100644 --- a/media/libvpx/libvpx/vpx_dsp/arm/highbd_vpx_convolve8_sve2.c +++ b/media/libvpx/libvpx/vpx_dsp/arm/highbd_vpx_convolve8_sve2.c @@ -15,6 +15,8 @@ #include "./vpx_dsp_rtcd.h" #include "vpx/vpx_integer.h" +#include "vpx_dsp/arm/highbd_convolve8_neon.h" +#include "vpx_dsp/arm/highbd_convolve8_sve.h" #include "vpx_dsp/arm/mem_neon.h" #include "vpx_dsp/arm/transpose_neon.h" #include "vpx_dsp/arm/vpx_neon_sve_bridge.h" @@ -31,6 +33,9 @@ DECLARE_ALIGNED(16, static const uint16_t, kDotProdMergeBlockTbl[24]) = { }; // clang-format on +DECLARE_ALIGNED(16, static const uint16_t, kTblConv4_8[8]) = { 0, 2, 4, 6, + 1, 3, 5, 7 }; + static INLINE void transpose_concat_4x4(const int16x4_t s0, const int16x4_t s1, const int16x4_t s2, const int16x4_t s3, int16x8_t res[2]) { @@ -450,3 +455,334 @@ void vpx_highbd_convolve8_avg_vert_sve2(const uint16_t *src, } while (w != 0); } } + +static INLINE void highbd_convolve_2d_4tap_sve2( + const uint16_t *src, ptrdiff_t src_stride, uint16_t *dst, + ptrdiff_t dst_stride, int w, int h, const int16x4_t x_filters, + const int16x4_t y_filters, int bd) { + const int16x8_t x_filter = vcombine_s16(x_filters, vdup_n_s16(0)); + + if (w == 4) { + const uint16x4_t max = vdup_n_u16((1 << bd) - 1); + const int16_t *s = (const int16_t *)src; + uint16_t *d = dst; + + int16x4_t h_s0[4], h_s1[4], h_s2[4]; + load_s16_4x4(s + 0 * src_stride, 1, &h_s0[0], &h_s0[1], &h_s0[2], &h_s0[3]); + load_s16_4x4(s + 1 * src_stride, 1, &h_s1[0], &h_s1[1], &h_s1[2], &h_s1[3]); + load_s16_4x4(s + 2 * src_stride, 1, &h_s2[0], &h_s2[1], &h_s2[2], &h_s2[3]); + + int16x4_t v_s0 = + vreinterpret_s16_u16(highbd_convolve4_4_sve(h_s0, x_filter, max)); + int16x4_t v_s1 = + vreinterpret_s16_u16(highbd_convolve4_4_sve(h_s1, x_filter, max)); + int16x4_t v_s2 = + vreinterpret_s16_u16(highbd_convolve4_4_sve(h_s2, x_filter, max)); + + s += 3 * src_stride; + + do { + int16x4_t h_s3[4], h_s4[4], h_s5[4], h_s6[4]; + load_s16_4x4(s + 0 * src_stride, 1, &h_s3[0], &h_s3[1], &h_s3[2], + &h_s3[3]); + load_s16_4x4(s + 1 * src_stride, 1, &h_s4[0], &h_s4[1], &h_s4[2], + &h_s4[3]); + load_s16_4x4(s + 2 * src_stride, 1, &h_s5[0], &h_s5[1], &h_s5[2], + &h_s5[3]); + load_s16_4x4(s + 3 * src_stride, 1, &h_s6[0], &h_s6[1], &h_s6[2], + &h_s6[3]); + + int16x4_t v_s3 = + vreinterpret_s16_u16(highbd_convolve4_4_sve(h_s3, x_filter, max)); + int16x4_t v_s4 = + vreinterpret_s16_u16(highbd_convolve4_4_sve(h_s4, x_filter, max)); + int16x4_t v_s5 = + vreinterpret_s16_u16(highbd_convolve4_4_sve(h_s5, x_filter, max)); + int16x4_t v_s6 = + vreinterpret_s16_u16(highbd_convolve4_4_sve(h_s6, x_filter, max)); + + uint16x4_t d0 = + highbd_convolve4_4_neon(v_s0, v_s1, v_s2, v_s3, y_filters, max); + uint16x4_t d1 = + highbd_convolve4_4_neon(v_s1, v_s2, v_s3, v_s4, y_filters, max); + uint16x4_t d2 = + highbd_convolve4_4_neon(v_s2, v_s3, v_s4, v_s5, y_filters, max); + uint16x4_t d3 = + highbd_convolve4_4_neon(v_s3, v_s4, v_s5, v_s6, y_filters, max); + + store_u16_4x4(d, dst_stride, d0, d1, d2, d3); + + v_s0 = v_s4; + v_s1 = v_s5; + v_s2 = v_s6; + s += 4 * src_stride; + d += 4 * dst_stride; + h -= 4; + } while (h != 0); + + } else { + const uint16x8_t max = vdupq_n_u16((1 << bd) - 1); + const uint16x8_t idx = vld1q_u16(kTblConv4_8); + + do { + const int16_t *s = (const int16_t *)src; + uint16_t *d = dst; + int height = h; + + int16x8_t h_s0[4], h_s1[4], h_s2[4]; + load_s16_8x4(s + 0 * src_stride, 1, &h_s0[0], &h_s0[1], &h_s0[2], + &h_s0[3]); + load_s16_8x4(s + 1 * src_stride, 1, &h_s1[0], &h_s1[1], &h_s1[2], + &h_s1[3]); + load_s16_8x4(s + 2 * src_stride, 1, &h_s2[0], &h_s2[1], &h_s2[2], + &h_s2[3]); + + int16x8_t v_s0 = vreinterpretq_s16_u16( + highbd_convolve4_8_sve(h_s0, x_filter, max, idx)); + int16x8_t v_s1 = vreinterpretq_s16_u16( + highbd_convolve4_8_sve(h_s1, x_filter, max, idx)); + int16x8_t v_s2 = vreinterpretq_s16_u16( + highbd_convolve4_8_sve(h_s2, x_filter, max, idx)); + + s += 3 * src_stride; + + do { + int16x8_t h_s3[4], h_s4[4], h_s5[4], h_s6[4]; + load_s16_8x4(s + 0 * src_stride, 1, &h_s3[0], &h_s3[1], &h_s3[2], + &h_s3[3]); + load_s16_8x4(s + 1 * src_stride, 1, &h_s4[0], &h_s4[1], &h_s4[2], + &h_s4[3]); + load_s16_8x4(s + 2 * src_stride, 1, &h_s5[0], &h_s5[1], &h_s5[2], + &h_s5[3]); + load_s16_8x4(s + 3 * src_stride, 1, &h_s6[0], &h_s6[1], &h_s6[2], + &h_s6[3]); + + int16x8_t v_s3 = vreinterpretq_s16_u16( + highbd_convolve4_8_sve(h_s3, x_filter, max, idx)); + int16x8_t v_s4 = vreinterpretq_s16_u16( + highbd_convolve4_8_sve(h_s4, x_filter, max, idx)); + int16x8_t v_s5 = vreinterpretq_s16_u16( + highbd_convolve4_8_sve(h_s5, x_filter, max, idx)); + int16x8_t v_s6 = vreinterpretq_s16_u16( + highbd_convolve4_8_sve(h_s6, x_filter, max, idx)); + + uint16x8_t d0 = + highbd_convolve4_8_neon(v_s0, v_s1, v_s2, v_s3, y_filters, max); + uint16x8_t d1 = + highbd_convolve4_8_neon(v_s1, v_s2, v_s3, v_s4, y_filters, max); + uint16x8_t d2 = + highbd_convolve4_8_neon(v_s2, v_s3, v_s4, v_s5, y_filters, max); + uint16x8_t d3 = + highbd_convolve4_8_neon(v_s3, v_s4, v_s5, v_s6, y_filters, max); + + store_u16_8x4(d, dst_stride, d0, d1, d2, d3); + + v_s0 = v_s4; + v_s1 = v_s5; + v_s2 = v_s6; + s += 4 * src_stride; + d += 4 * dst_stride; + height -= 4; + } while (height != 0); + src += 8; + dst += 8; + w -= 8; + } while (w != 0); + } +} + +static INLINE void highbd_convolve8_2d_horiz_sve2( + const uint16_t *src, ptrdiff_t src_stride, uint16_t *dst, + ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, + int y0_q4, int y_step_q4, int w, int h, int bd) { + assert((intptr_t)dst % 4 == 0); + assert(dst_stride % 4 == 0); + assert(x_step_q4 == 16); + assert(h % 4 == 3 && h >= 7); + + (void)x_step_q4; + (void)y0_q4; + (void)y_step_q4; + + const int16x8_t filters = vld1q_s16(filter[x0_q4]); + + src -= 3; + + if (w == 4) { + const uint16x4_t max = vdup_n_u16((1 << bd) - 1); + const int16_t *s = (const int16_t *)src; + uint16_t *d = dst; + + do { + int16x8_t s0[4], s1[4], s2[4], s3[4]; + load_s16_8x4(s + 0 * src_stride, 1, &s0[0], &s0[1], &s0[2], &s0[3]); + load_s16_8x4(s + 1 * src_stride, 1, &s1[0], &s1[1], &s1[2], &s1[3]); + load_s16_8x4(s + 2 * src_stride, 1, &s2[0], &s2[1], &s2[2], &s2[3]); + load_s16_8x4(s + 3 * src_stride, 1, &s3[0], &s3[1], &s3[2], &s3[3]); + + uint16x4_t d0 = highbd_convolve8_4(s0, filters, max); + uint16x4_t d1 = highbd_convolve8_4(s1, filters, max); + uint16x4_t d2 = highbd_convolve8_4(s2, filters, max); + uint16x4_t d3 = highbd_convolve8_4(s3, filters, max); + + store_u16_4x4(d, dst_stride, d0, d1, d2, d3); + + s += 4 * src_stride; + d += 4 * dst_stride; + h -= 4; + } while (h != 3); + + // Process final three rows (h % 4 == 3). + int16x8_t s0[4], s1[4], s2[4]; + load_s16_8x4(s + 0 * src_stride, 1, &s0[0], &s0[1], &s0[2], &s0[3]); + load_s16_8x4(s + 1 * src_stride, 1, &s1[0], &s1[1], &s1[2], &s1[3]); + load_s16_8x4(s + 2 * src_stride, 1, &s2[0], &s2[1], &s2[2], &s2[3]); + + uint16x4_t d0 = highbd_convolve8_4(s0, filters, max); + uint16x4_t d1 = highbd_convolve8_4(s1, filters, max); + uint16x4_t d2 = highbd_convolve8_4(s2, filters, max); + + store_u16_4x3(d, dst_stride, d0, d1, d2); + } else { + const uint16x8_t max = vdupq_n_u16((1 << bd) - 1); + + do { + const int16_t *s = (const int16_t *)src; + uint16_t *d = dst; + int width = w; + + do { + int16x8_t s0[8], s1[8], s2[8], s3[8]; + load_s16_8x8(s + 0 * src_stride, 1, &s0[0], &s0[1], &s0[2], &s0[3], + &s0[4], &s0[5], &s0[6], &s0[7]); + load_s16_8x8(s + 1 * src_stride, 1, &s1[0], &s1[1], &s1[2], &s1[3], + &s1[4], &s1[5], &s1[6], &s1[7]); + load_s16_8x8(s + 2 * src_stride, 1, &s2[0], &s2[1], &s2[2], &s2[3], + &s2[4], &s2[5], &s2[6], &s2[7]); + load_s16_8x8(s + 3 * src_stride, 1, &s3[0], &s3[1], &s3[2], &s3[3], + &s3[4], &s3[5], &s3[6], &s3[7]); + + uint16x8_t d0 = highbd_convolve8_8(s0, filters, max); + uint16x8_t d1 = highbd_convolve8_8(s1, filters, max); + uint16x8_t d2 = highbd_convolve8_8(s2, filters, max); + uint16x8_t d3 = highbd_convolve8_8(s3, filters, max); + + store_u16_8x4(d, dst_stride, d0, d1, d2, d3); + + s += 8; + d += 8; + width -= 8; + } while (width != 0); + src += 4 * src_stride; + dst += 4 * dst_stride; + h -= 4; + } while (h != 3); + + // Process final three rows (h % 4 == 3). + const int16_t *s = (const int16_t *)src; + uint16_t *d = dst; + int width = w; + + do { + int16x8_t s0[8], s1[8], s2[8]; + load_s16_8x8(s + 0 * src_stride, 1, &s0[0], &s0[1], &s0[2], &s0[3], + &s0[4], &s0[5], &s0[6], &s0[7]); + load_s16_8x8(s + 1 * src_stride, 1, &s1[0], &s1[1], &s1[2], &s1[3], + &s1[4], &s1[5], &s1[6], &s1[7]); + load_s16_8x8(s + 2 * src_stride, 1, &s2[0], &s2[1], &s2[2], &s2[3], + &s2[4], &s2[5], &s2[6], &s2[7]); + + uint16x8_t d0 = highbd_convolve8_8(s0, filters, max); + uint16x8_t d1 = highbd_convolve8_8(s1, filters, max); + uint16x8_t d2 = highbd_convolve8_8(s2, filters, max); + + store_u16_8x3(d, dst_stride, d0, d1, d2); + + s += 8; + d += 8; + width -= 8; + } while (width != 0); + } +} + +void vpx_highbd_convolve8_sve2(const uint16_t *src, ptrdiff_t src_stride, + uint16_t *dst, ptrdiff_t dst_stride, + const InterpKernel *filter, int x0_q4, + int x_step_q4, int y0_q4, int y_step_q4, int w, + int h, int bd) { + if (x_step_q4 != 16 || y_step_q4 != 16) { + vpx_highbd_convolve8_c(src, src_stride, dst, dst_stride, filter, x0_q4, + x_step_q4, y0_q4, y_step_q4, w, h, bd); + return; + } + + assert(y_step_q4 == 16); + assert(x_step_q4 == 16); + + const int horiz_filter_taps = vpx_get_filter_taps(filter[x0_q4]) <= 4 ? 4 : 8; + const int vert_filter_taps = vpx_get_filter_taps(filter[y0_q4]) <= 4 ? 4 : 8; + + if (horiz_filter_taps == 4 || vert_filter_taps == 4) { + const ptrdiff_t horiz_offset = horiz_filter_taps / 2 - 1; + const ptrdiff_t vert_offset = (vert_filter_taps / 2 - 1) * src_stride; + const int16x4_t x_filter = vld1_s16(filter[x0_q4] + 2); + const int16x4_t y_filter = vld1_s16(filter[y0_q4] + 2); + + highbd_convolve_2d_4tap_sve2(src - horiz_offset - vert_offset, src_stride, + dst, dst_stride, w, h, x_filter, y_filter, bd); + return; + } + + // Given our constraints: w <= 64, h <= 64, taps <= 8 we can reduce the + // maximum buffer size to 64 * (64 + 7). + DECLARE_ALIGNED(32, uint16_t, im_block[64 * 71]); + const int im_stride = 64; + + // Account for the vertical phase needing SUBPEL_TAPS / 2 - 1 lines prior + // and SUBPEL_TAPS / 2 lines post. + const int im_height = h + SUBPEL_TAPS - 1; + const ptrdiff_t border_offset = SUBPEL_TAPS / 2 - 1; + + highbd_convolve8_2d_horiz_sve2(src - src_stride * border_offset, src_stride, + im_block, im_stride, filter, x0_q4, x_step_q4, + y0_q4, y_step_q4, w, im_height, bd); + + // Step into the temporary buffer border_offset rows to get actual frame data. + vpx_highbd_convolve8_vert_sve2(im_block + im_stride * border_offset, + im_stride, dst, dst_stride, filter, x0_q4, + x_step_q4, y0_q4, y_step_q4, w, h, bd); +} + +void vpx_highbd_convolve8_avg_sve2(const uint16_t *src, ptrdiff_t src_stride, + uint16_t *dst, ptrdiff_t dst_stride, + const InterpKernel *filter, int x0_q4, + int x_step_q4, int y0_q4, int y_step_q4, + int w, int h, int bd) { + if (x_step_q4 != 16 || y_step_q4 != 16) { + vpx_highbd_convolve8_c(src, src_stride, dst, dst_stride, filter, x0_q4, + x_step_q4, y0_q4, y_step_q4, w, h, bd); + return; + } + + assert(y_step_q4 == 16); + assert(x_step_q4 == 16); + + // Given our constraints: w <= 64, h <= 64, taps <= 8 we can reduce the + // maximum buffer size to 64 * (64 + 7). + DECLARE_ALIGNED(32, uint16_t, im_block[64 * 71]); + const int im_stride = 64; + + // Account for the vertical phase needing SUBPEL_TAPS / 2 - 1 lines prior + // and SUBPEL_TAPS / 2 lines post. + const int im_height = h + SUBPEL_TAPS - 1; + const ptrdiff_t border_offset = SUBPEL_TAPS / 2 - 1; + + highbd_convolve8_2d_horiz_sve2(src - src_stride * border_offset, src_stride, + im_block, im_stride, filter, x0_q4, x_step_q4, + y0_q4, y_step_q4, w, im_height, bd); + + // Step into the temporary buffer border_offset rows to get actual frame data. + vpx_highbd_convolve8_avg_vert_sve2(im_block + im_stride * border_offset, + im_stride, dst, dst_stride, filter, x0_q4, + x_step_q4, y0_q4, y_step_q4, w, h, bd); +} diff --git a/media/libvpx/libvpx/vpx_dsp/arm/vpx_scaled_convolve8_neon.c b/media/libvpx/libvpx/vpx_dsp/arm/vpx_scaled_convolve8_neon.c index 9bd5ec285c..f40b6a907f 100644 --- a/media/libvpx/libvpx/vpx_dsp/arm/vpx_scaled_convolve8_neon.c +++ b/media/libvpx/libvpx/vpx_dsp/arm/vpx_scaled_convolve8_neon.c @@ -18,6 +18,7 @@ #include "vpx_dsp/arm/mem_neon.h" #include "vpx_dsp/arm/transpose_neon.h" #include "vpx_dsp/arm/vpx_convolve8_neon.h" +#include "vpx_dsp/vpx_filter.h" #include "vpx_ports/mem.h" static INLINE void scaledconvolve_horiz_neon( diff --git a/media/libvpx/libvpx/vpx_dsp/bitwriter.c b/media/libvpx/libvpx/vpx_dsp/bitwriter.c index 5b41aa54dd..d3ef9bd89a 100644 --- a/media/libvpx/libvpx/vpx_dsp/bitwriter.c +++ b/media/libvpx/libvpx/vpx_dsp/bitwriter.c @@ -9,6 +9,7 @@ */ #include <assert.h> +#include <limits.h> #include "./bitwriter.h" @@ -16,16 +17,20 @@ #include "vpx_util/vpx_debug_util.h" #endif -void vpx_start_encode(vpx_writer *br, uint8_t *source) { +void vpx_start_encode(vpx_writer *br, uint8_t *source, size_t size) { br->lowvalue = 0; br->range = 255; br->count = -24; - br->buffer = source; + br->error = 0; br->pos = 0; + // Make sure it is safe to cast br->pos to int in vpx_write(). + if (size > INT_MAX) size = INT_MAX; + br->size = (unsigned int)size; + br->buffer = source; vpx_write_bit(br, 0); } -void vpx_stop_encode(vpx_writer *br) { +int vpx_stop_encode(vpx_writer *br) { int i; #if CONFIG_BITSTREAM_DEBUG @@ -34,9 +39,17 @@ void vpx_stop_encode(vpx_writer *br) { for (i = 0; i < 32; i++) vpx_write_bit(br, 0); // Ensure there's no ambigous collision with any index marker bytes - if ((br->buffer[br->pos - 1] & 0xe0) == 0xc0) br->buffer[br->pos++] = 0; + if (!br->error && (br->buffer[br->pos - 1] & 0xe0) == 0xc0) { + if (br->pos < br->size) { + br->buffer[br->pos++] = 0; + } else { + br->error = 1; + } + } #if CONFIG_BITSTREAM_DEBUG bitstream_queue_set_skip_write(0); #endif + + return br->error ? -1 : 0; } diff --git a/media/libvpx/libvpx/vpx_dsp/bitwriter.h b/media/libvpx/libvpx/vpx_dsp/bitwriter.h index 5f1ee69ec2..daff331daf 100644 --- a/media/libvpx/libvpx/vpx_dsp/bitwriter.h +++ b/media/libvpx/libvpx/vpx_dsp/bitwriter.h @@ -29,12 +29,19 @@ typedef struct vpx_writer { unsigned int lowvalue; unsigned int range; int count; + // Whether there has been an error. + int error; + // We maintain the invariant that pos <= size, i.e., we never write beyond + // the end of the buffer. If pos would be incremented to be greater than + // size, leave pos unchanged and set error to 1. unsigned int pos; + unsigned int size; uint8_t *buffer; } vpx_writer; -void vpx_start_encode(vpx_writer *br, uint8_t *source); -void vpx_stop_encode(vpx_writer *br); +void vpx_start_encode(vpx_writer *br, uint8_t *source, size_t size); +// Returns 0 on success and returns -1 in case of error. +int vpx_stop_encode(vpx_writer *br); static INLINE VPX_NO_UNSIGNED_SHIFT_CHECK void vpx_write(vpx_writer *br, int bit, @@ -77,18 +84,25 @@ static INLINE VPX_NO_UNSIGNED_SHIFT_CHECK void vpx_write(vpx_writer *br, if (count >= 0) { int offset = shift - count; - if ((lowvalue << (offset - 1)) & 0x80000000) { - int x = br->pos - 1; + if (!br->error) { + if ((lowvalue << (offset - 1)) & 0x80000000) { + int x = (int)br->pos - 1; - while (x >= 0 && br->buffer[x] == 0xff) { - br->buffer[x] = 0; - x--; + while (x >= 0 && br->buffer[x] == 0xff) { + br->buffer[x] = 0; + x--; + } + + // TODO(wtc): How to prove x >= 0? + br->buffer[x] += 1; } - br->buffer[x] += 1; + if (br->pos < br->size) { + br->buffer[br->pos++] = (lowvalue >> (24 - offset)) & 0xff; + } else { + br->error = 1; + } } - - br->buffer[br->pos++] = (lowvalue >> (24 - offset)) & 0xff; lowvalue <<= offset; shift = count; lowvalue &= 0xffffff; diff --git a/media/libvpx/libvpx/vpx_dsp/bitwriter_buffer.c b/media/libvpx/libvpx/vpx_dsp/bitwriter_buffer.c index 7a7e96f02e..b3a2490f2c 100644 --- a/media/libvpx/libvpx/vpx_dsp/bitwriter_buffer.c +++ b/media/libvpx/libvpx/vpx_dsp/bitwriter_buffer.c @@ -8,24 +8,43 @@ * be found in the AUTHORS file in the root of the source tree. */ +#include <assert.h> #include <limits.h> #include <stdlib.h> #include "./vpx_config.h" #include "./bitwriter_buffer.h" +void vpx_wb_init(struct vpx_write_bit_buffer *wb, uint8_t *bit_buffer, + size_t size) { + wb->error = 0; + wb->bit_offset = 0; + wb->size = size; + wb->bit_buffer = bit_buffer; +} + +int vpx_wb_has_error(const struct vpx_write_bit_buffer *wb) { + return wb->error; +} + size_t vpx_wb_bytes_written(const struct vpx_write_bit_buffer *wb) { + assert(!wb->error); return wb->bit_offset / CHAR_BIT + (wb->bit_offset % CHAR_BIT > 0); } void vpx_wb_write_bit(struct vpx_write_bit_buffer *wb, int bit) { + if (wb->error) return; const int off = (int)wb->bit_offset; const int p = off / CHAR_BIT; const int q = CHAR_BIT - 1 - off % CHAR_BIT; + if ((size_t)p >= wb->size) { + wb->error = 1; + return; + } if (q == CHAR_BIT - 1) { wb->bit_buffer[p] = bit << q; } else { - wb->bit_buffer[p] &= ~(1 << q); + assert((wb->bit_buffer[p] & (1 << q)) == 0); wb->bit_buffer[p] |= bit << q; } wb->bit_offset = off + 1; diff --git a/media/libvpx/libvpx/vpx_dsp/bitwriter_buffer.h b/media/libvpx/libvpx/vpx_dsp/bitwriter_buffer.h index 3662cb64df..3ee0e9658b 100644 --- a/media/libvpx/libvpx/vpx_dsp/bitwriter_buffer.h +++ b/media/libvpx/libvpx/vpx_dsp/bitwriter_buffer.h @@ -18,10 +18,24 @@ extern "C" { #endif struct vpx_write_bit_buffer { - uint8_t *bit_buffer; + // Whether there has been an error. + int error; + // We maintain the invariant that bit_offset <= size * CHAR_BIT, i.e., we + // never write beyond the end of bit_buffer. If bit_offset would be + // incremented to be greater than size * CHAR_BIT, leave bit_offset unchanged + // and set error to 1. size_t bit_offset; + // Size of bit_buffer in bytes. + size_t size; + uint8_t *bit_buffer; }; +void vpx_wb_init(struct vpx_write_bit_buffer *wb, uint8_t *bit_buffer, + size_t size); + +int vpx_wb_has_error(const struct vpx_write_bit_buffer *wb); + +// Must not be called if vpx_wb_has_error(wb) returns true. size_t vpx_wb_bytes_written(const struct vpx_write_bit_buffer *wb); void vpx_wb_write_bit(struct vpx_write_bit_buffer *wb, int bit); diff --git a/media/libvpx/libvpx/vpx_dsp/vpx_dsp_rtcd_defs.pl b/media/libvpx/libvpx/vpx_dsp/vpx_dsp_rtcd_defs.pl index f40f85c036..d08f44ac1d 100644 --- a/media/libvpx/libvpx/vpx_dsp/vpx_dsp_rtcd_defs.pl +++ b/media/libvpx/libvpx/vpx_dsp/vpx_dsp_rtcd_defs.pl @@ -424,7 +424,7 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") { specialize qw/vpx_highbd_convolve_avg sse2 avx2 neon/; add_proto qw/void vpx_highbd_convolve8/, "const uint16_t *src, ptrdiff_t src_stride, uint16_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h, int bd"; - specialize qw/vpx_highbd_convolve8 avx2 neon/, "$sse2_x86_64"; + specialize qw/vpx_highbd_convolve8 avx2 neon sve2/, "$sse2_x86_64"; add_proto qw/void vpx_highbd_convolve8_horiz/, "const uint16_t *src, ptrdiff_t src_stride, uint16_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h, int bd"; specialize qw/vpx_highbd_convolve8_horiz avx2 neon sve/, "$sse2_x86_64"; @@ -433,7 +433,7 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") { specialize qw/vpx_highbd_convolve8_vert avx2 neon sve2/, "$sse2_x86_64"; add_proto qw/void vpx_highbd_convolve8_avg/, "const uint16_t *src, ptrdiff_t src_stride, uint16_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h, int bd"; - specialize qw/vpx_highbd_convolve8_avg avx2 neon/, "$sse2_x86_64"; + specialize qw/vpx_highbd_convolve8_avg avx2 neon sve2/, "$sse2_x86_64"; add_proto qw/void vpx_highbd_convolve8_avg_horiz/, "const uint16_t *src, ptrdiff_t src_stride, uint16_t *dst, ptrdiff_t dst_stride, const InterpKernel *filter, int x0_q4, int x_step_q4, int y0_q4, int y_step_q4, int w, int h, int bd"; specialize qw/vpx_highbd_convolve8_avg_horiz avx2 neon sve/, "$sse2_x86_64"; diff --git a/media/libvpx/libvpx/vpx_ports/aarch64_cpudetect.c b/media/libvpx/libvpx/vpx_ports/aarch64_cpudetect.c index eba12d312a..5c4c16a390 100644 --- a/media/libvpx/libvpx/vpx_ports/aarch64_cpudetect.c +++ b/media/libvpx/libvpx/vpx_ports/aarch64_cpudetect.c @@ -9,7 +9,8 @@ */ #include "./vpx_config.h" -#include "arm_cpudetect.h" +#include "vpx_ports/arm.h" +#include "vpx_ports/arm_cpudetect.h" #if defined(__APPLE__) #include <sys/sysctl.h> diff --git a/media/libvpx/libvpx/vpx_util/vpx_thread.h b/media/libvpx/libvpx/vpx_util/vpx_thread.h index 11a1d74387..22051eae41 100644 --- a/media/libvpx/libvpx/vpx_util/vpx_thread.h +++ b/media/libvpx/libvpx/vpx_util/vpx_thread.h @@ -19,8 +19,6 @@ extern "C" { #endif -#define MAX_NUM_THREADS 64 - // State of the worker thread object typedef enum { VPX_WORKER_STATUS_NOT_OK = 0, // object is unusable diff --git a/media/libvpx/missing_header.patch b/media/libvpx/missing_header.patch deleted file mode 100644 index 02b77170ee..0000000000 --- a/media/libvpx/missing_header.patch +++ /dev/null @@ -1,12 +0,0 @@ -Add missing header for EBUSY - ---- a/vpx_util/vpx_pthread.h -+++ b/vpx_util/vpx_pthread.h -@@ -26,6 +26,7 @@ extern "C" { - #define NOMINMAX - #undef WIN32_LEAN_AND_MEAN - #define WIN32_LEAN_AND_MEAN -+#include <errno.h> // NOLINT - #include <process.h> // NOLINT - #include <stddef.h> // NOLINT - #include <windows.h> // NOLINT diff --git a/media/libvpx/moz.build b/media/libvpx/moz.build index 635b5d0fdd..df29c9aae8 100644 --- a/media/libvpx/moz.build +++ b/media/libvpx/moz.build @@ -14,45 +14,46 @@ if CONFIG['VPX_USE_NASM']: # Linux, Mac and Win share file lists for x86* but not configurations. if CONFIG['TARGET_CPU'] == 'x86_64': - EXPORTS.vpx += files['X64_EXPORTS'] - SOURCES += files['X64_SOURCES'] if CONFIG['OS_TARGET'] == 'WINNT': + EXPORTS.vpx += files['WIN_X64_EXPORTS'] + SOURCES += files['WIN_X64_SOURCES'] ASFLAGS += [ '-I%s/media/libvpx/config/win/x64/' % TOPSRCDIR ] LOCAL_INCLUDES += [ '/media/libvpx/config/win/x64/' ] - SOURCES += [ '/media/libvpx/config/win/x64/vpx_config.c' ] elif CONFIG['OS_TARGET'] == 'Darwin': + EXPORTS.vpx += files['MAC_X64_EXPORTS'] + SOURCES += files['MAC_X64_SOURCES'] ASFLAGS += [ '-I%s/media/libvpx/config/mac/x64/' % TOPSRCDIR ] LOCAL_INCLUDES += [ '/media/libvpx/config/mac/x64/' ] - SOURCES += [ '/media/libvpx/config/mac/x64/vpx_config.c' ] else: # Android, Linux, BSDs, etc. + EXPORTS.vpx += files['LINUX_X64_EXPORTS'] + SOURCES += files['LINUX_X64_SOURCES'] ASFLAGS += [ '-I%s/media/libvpx/config/linux/x64/' % TOPSRCDIR ] LOCAL_INCLUDES += [ '/media/libvpx/config/linux/x64/' ] - SOURCES += [ '/media/libvpx/config/linux/x64/vpx_config.c' ] elif CONFIG['TARGET_CPU'] == 'x86': - EXPORTS.vpx += files['IA32_EXPORTS'] - SOURCES += files['IA32_SOURCES'] if CONFIG['OS_TARGET'] == 'WINNT': + EXPORTS.vpx += files['WIN_IA32_EXPORTS'] + SOURCES += files['WIN_IA32_SOURCES'] ASFLAGS += [ '-I%s/media/libvpx/config/win/ia32/' % TOPSRCDIR ] LOCAL_INCLUDES += [ '/media/libvpx/config/win/ia32/' ] - SOURCES += [ '/media/libvpx/config/win/ia32/vpx_config.c' ] elif CONFIG['OS_TARGET'] == 'Darwin': + EXPORTS.vpx += files['MAC_IA32_EXPORTS'] + SOURCES += files['MAC_IA32_SOURCES'] ASFLAGS += [ '-I%s/media/libvpx/config/mac/ia32/' % TOPSRCDIR ] LOCAL_INCLUDES += [ '/media/libvpx/config/mac/ia32/' ] - SOURCES += [ '/media/libvpx/config/mac/ia32/vpx_config.c' ] else: # Android, Linux, BSDs, etc. + EXPORTS.vpx += files['LINUX_IA32_EXPORTS'] + SOURCES += files['LINUX_IA32_SOURCES'] ASFLAGS += [ '-I%s/media/libvpx/config/linux/ia32/' % TOPSRCDIR ] LOCAL_INCLUDES += [ '/media/libvpx/config/linux/ia32/' ] - SOURCES += [ '/media/libvpx/config/linux/ia32/vpx_config.c' ] elif CONFIG['TARGET_CPU'] == 'arm': - EXPORTS.vpx += files['ARM_EXPORTS'] + EXPORTS.vpx += files['LINUX_ARM_EXPORTS'] ASFLAGS += [ '-I%s/media/libvpx/config/linux/arm/' % TOPSRCDIR, '-I%s/libvpx' % OBJDIR, ] LOCAL_INCLUDES += [ '/media/libvpx/config/linux/arm/' ] - SOURCES += [ '/media/libvpx/config/linux/arm/vpx_config.c' ] - arm_asm_files = files['ARM_SOURCES'] + arm_asm_files = files['LINUX_ARM_SOURCES'] if CONFIG['GNU_AS']: SOURCES += sorted([ @@ -70,28 +71,28 @@ elif CONFIG['TARGET_CPU'] == 'arm': LOCAL_INCLUDES += [ '%%%s/sources/android/cpufeatures' % CONFIG['ANDROID_NDK'], ] -elif CONFIG['TARGET_CPU'] == 'aarch64' and CONFIG['OS_TARGET'] == 'WINNT': - EXPORTS.vpx += files['ARM64_EXPORTS'] - # Bug 1885585: clang on win/aarch64 cannot compile SVInt8_t type for now. - SOURCES += [ - f for f in files['ARM64_SOURCES'] if not f.endswith('_sve.c') - ] - ASFLAGS += [ '-I%s/media/libvpx/config/win/aarch64/' % TOPSRCDIR ] - LOCAL_INCLUDES += [ '/media/libvpx/config/win/aarch64/' ] - SOURCES += [ '/media/libvpx/config/win/aarch64/vpx_config.c' ] elif CONFIG['TARGET_CPU'] == 'aarch64': - EXPORTS.vpx += files['ARM64_EXPORTS'] - SOURCES += files['ARM64_SOURCES'] - ASFLAGS += [ '-I%s/media/libvpx/config/linux/arm64/' % TOPSRCDIR ] - LOCAL_INCLUDES += [ '/media/libvpx/config/linux/arm64/' ] - SOURCES += [ '/media/libvpx/config/linux/arm64/vpx_config.c' ] + if CONFIG['OS_TARGET'] == 'WINNT': + EXPORTS.vpx += files['WIN_AARCH64_EXPORTS'] + SOURCES += files['WIN_AARCH64_SOURCES'] + ASFLAGS += [ '-I%s/media/libvpx/config/win/aarch64/' % TOPSRCDIR ] + LOCAL_INCLUDES += [ '/media/libvpx/config/win/aarch64/' ] + elif CONFIG['OS_TARGET'] == 'Darwin': + EXPORTS.vpx += files['MAC_ARM64_EXPORTS'] + SOURCES += files['MAC_ARM64_SOURCES'] + ASFLAGS += [ '-I%s/media/libvpx/config/mac/arm64/' % TOPSRCDIR ] + LOCAL_INCLUDES += [ '/media/libvpx/config/mac/arm64/' ] + else: # Android, Linux, BSDs, etc. + EXPORTS.vpx += files['LINUX_ARM64_EXPORTS'] + SOURCES += files['LINUX_ARM64_SOURCES'] + ASFLAGS += [ '-I%s/media/libvpx/config/linux/arm64/' % TOPSRCDIR ] + LOCAL_INCLUDES += [ '/media/libvpx/config/linux/arm64/' ] else: # Generic C-only configuration EXPORTS.vpx += files['GENERIC_EXPORTS'] SOURCES += files['GENERIC_SOURCES'] ASFLAGS += [ '-I%s/media/libvpx/config/generic/' % TOPSRCDIR ] LOCAL_INCLUDES += [ '/media/libvpx/config/generic/' ] - SOURCES += [ '/media/libvpx/config/generic/vpx_config.c' ] # We allow warnings for third-party code that can be updated from upstream. AllowCompilerWarnings() diff --git a/media/libvpx/moz.yaml b/media/libvpx/moz.yaml index 0b3ec52482..15f1398932 100644 --- a/media/libvpx/moz.yaml +++ b/media/libvpx/moz.yaml @@ -20,11 +20,11 @@ origin: # Human-readable identifier for this version/release # Generally "version NNN", "tag SSS", "bookmark SSS" - release: 7fb8ceccf92c35cd5131b05c0502916715ebc76b (Fri Mar 15 01:11:50 2024). + release: 85dafa9c61f99330f484e77297684b42af6ff37d (Tue Apr 16 18:51:27 2024). # Revision to pull in # Must be a long or short commit SHA (long preferred) - revision: 7fb8ceccf92c35cd5131b05c0502916715ebc76b + revision: 85dafa9c61f99330f484e77297684b42af6ff37d # The package's license, where possible using the mnemonic from # https://spdx.org/licenses/ @@ -56,7 +56,6 @@ vendoring: - arm_cpu_runtime_detection_code_on_openbsd.patch - input_frame_validation.patch - input_frame_validation_vp9.patch - - missing_header.patch update-actions: - action: move-file diff --git a/media/libvpx/sources.mozbuild b/media/libvpx/sources.mozbuild index 1ad5d4447c..12c20ede9f 100644 --- a/media/libvpx/sources.mozbuild +++ b/media/libvpx/sources.mozbuild @@ -1,7 +1,7 @@ # This file is generated. Do not edit. files = { - 'X64_EXPORTS': [ + 'LINUX_X64_EXPORTS': [ 'libvpx/vpx/vp8.h', 'libvpx/vpx/vp8cx.h', 'libvpx/vpx/vp8dx.h', @@ -25,7 +25,8 @@ files = { 'libvpx/vpx_scale/vpx_scale.h', 'libvpx/vpx_scale/yv12config.h', ], - 'X64_SOURCES': [ + 'LINUX_X64_SOURCES': [ + 'libvpx/../config/linux/x64/vpx_config.c', 'libvpx/vp8/common/alloccommon.c', 'libvpx/vp8/common/blockd.c', 'libvpx/vp8/common/dequantize.c', @@ -200,7 +201,6 @@ files = { 'libvpx/vpx/src/vpx_decoder.c', 'libvpx/vpx/src/vpx_encoder.c', 'libvpx/vpx/src/vpx_image.c', - 'libvpx/vpx/src/vpx_tpl.c', 'libvpx/vpx_dsp/add_noise.c', 'libvpx/vpx_dsp/avg.c', 'libvpx/vpx_dsp/bitreader.c', @@ -278,7 +278,7 @@ files = { 'libvpx/vpx_util/vpx_thread.c', 'libvpx/vpx_util/vpx_write_yuv_frame.c', ], - 'IA32_EXPORTS': [ + 'MAC_X64_EXPORTS': [ 'libvpx/vpx/vp8.h', 'libvpx/vpx/vp8cx.h', 'libvpx/vpx/vp8dx.h', @@ -302,7 +302,562 @@ files = { 'libvpx/vpx_scale/vpx_scale.h', 'libvpx/vpx_scale/yv12config.h', ], - 'IA32_SOURCES': [ + 'MAC_X64_SOURCES': [ + 'libvpx/../config/mac/x64/vpx_config.c', + 'libvpx/vp8/common/alloccommon.c', + 'libvpx/vp8/common/blockd.c', + 'libvpx/vp8/common/dequantize.c', + 'libvpx/vp8/common/entropy.c', + 'libvpx/vp8/common/entropymode.c', + 'libvpx/vp8/common/entropymv.c', + 'libvpx/vp8/common/extend.c', + 'libvpx/vp8/common/filter.c', + 'libvpx/vp8/common/findnearmv.c', + 'libvpx/vp8/common/generic/systemdependent.c', + 'libvpx/vp8/common/idct_blk.c', + 'libvpx/vp8/common/idctllm.c', + 'libvpx/vp8/common/loopfilter_filters.c', + 'libvpx/vp8/common/mbpitch.c', + 'libvpx/vp8/common/mfqe.c', + 'libvpx/vp8/common/modecont.c', + 'libvpx/vp8/common/postproc.c', + 'libvpx/vp8/common/quant_common.c', + 'libvpx/vp8/common/reconinter.c', + 'libvpx/vp8/common/reconintra.c', + 'libvpx/vp8/common/reconintra4x4.c', + 'libvpx/vp8/common/rtcd.c', + 'libvpx/vp8/common/setupintrarecon.c', + 'libvpx/vp8/common/swapyv12buffer.c', + 'libvpx/vp8/common/treecoder.c', + 'libvpx/vp8/common/vp8_loopfilter.c', + 'libvpx/vp8/common/vp8_skin_detection.c', + 'libvpx/vp8/common/x86/bilinear_filter_sse2.c', + 'libvpx/vp8/common/x86/dequantize_mmx.asm', + 'libvpx/vp8/common/x86/idct_blk_mmx.c', + 'libvpx/vp8/common/x86/idct_blk_sse2.c', + 'libvpx/vp8/common/x86/idctllm_mmx.asm', + 'libvpx/vp8/common/x86/idctllm_sse2.asm', + 'libvpx/vp8/common/x86/iwalsh_sse2.asm', + 'libvpx/vp8/common/x86/loopfilter_block_sse2_x86_64.asm', + 'libvpx/vp8/common/x86/loopfilter_sse2.asm', + 'libvpx/vp8/common/x86/loopfilter_x86.c', + 'libvpx/vp8/common/x86/mfqe_sse2.asm', + 'libvpx/vp8/common/x86/recon_mmx.asm', + 'libvpx/vp8/common/x86/recon_sse2.asm', + 'libvpx/vp8/common/x86/subpixel_mmx.asm', + 'libvpx/vp8/common/x86/subpixel_sse2.asm', + 'libvpx/vp8/common/x86/subpixel_ssse3.asm', + 'libvpx/vp8/common/x86/vp8_asm_stubs.c', + 'libvpx/vp8/decoder/dboolhuff.c', + 'libvpx/vp8/decoder/decodeframe.c', + 'libvpx/vp8/decoder/decodemv.c', + 'libvpx/vp8/decoder/detokenize.c', + 'libvpx/vp8/decoder/onyxd_if.c', + 'libvpx/vp8/decoder/threading.c', + 'libvpx/vp8/encoder/bitstream.c', + 'libvpx/vp8/encoder/boolhuff.c', + 'libvpx/vp8/encoder/copy_c.c', + 'libvpx/vp8/encoder/dct.c', + 'libvpx/vp8/encoder/denoising.c', + 'libvpx/vp8/encoder/encodeframe.c', + 'libvpx/vp8/encoder/encodeintra.c', + 'libvpx/vp8/encoder/encodemb.c', + 'libvpx/vp8/encoder/encodemv.c', + 'libvpx/vp8/encoder/ethreading.c', + 'libvpx/vp8/encoder/firstpass.c', + 'libvpx/vp8/encoder/lookahead.c', + 'libvpx/vp8/encoder/mcomp.c', + 'libvpx/vp8/encoder/modecosts.c', + 'libvpx/vp8/encoder/mr_dissim.c', + 'libvpx/vp8/encoder/onyx_if.c', + 'libvpx/vp8/encoder/pickinter.c', + 'libvpx/vp8/encoder/picklpf.c', + 'libvpx/vp8/encoder/ratectrl.c', + 'libvpx/vp8/encoder/rdopt.c', + 'libvpx/vp8/encoder/segmentation.c', + 'libvpx/vp8/encoder/temporal_filter.c', + 'libvpx/vp8/encoder/tokenize.c', + 'libvpx/vp8/encoder/treewriter.c', + 'libvpx/vp8/encoder/vp8_quantize.c', + 'libvpx/vp8/encoder/x86/block_error_sse2.asm', + 'libvpx/vp8/encoder/x86/copy_sse2.asm', + 'libvpx/vp8/encoder/x86/copy_sse3.asm', + 'libvpx/vp8/encoder/x86/dct_sse2.asm', + 'libvpx/vp8/encoder/x86/denoising_sse2.c', + 'libvpx/vp8/encoder/x86/fwalsh_sse2.asm', + 'libvpx/vp8/encoder/x86/quantize_sse4.c', + 'libvpx/vp8/encoder/x86/temporal_filter_apply_sse2.asm', + 'libvpx/vp8/encoder/x86/vp8_enc_stubs_sse2.c', + 'libvpx/vp8/encoder/x86/vp8_quantize_sse2.c', + 'libvpx/vp8/encoder/x86/vp8_quantize_ssse3.c', + 'libvpx/vp8/vp8_cx_iface.c', + 'libvpx/vp8/vp8_dx_iface.c', + 'libvpx/vp9/common/vp9_alloccommon.c', + 'libvpx/vp9/common/vp9_blockd.c', + 'libvpx/vp9/common/vp9_common_data.c', + 'libvpx/vp9/common/vp9_entropy.c', + 'libvpx/vp9/common/vp9_entropymode.c', + 'libvpx/vp9/common/vp9_entropymv.c', + 'libvpx/vp9/common/vp9_filter.c', + 'libvpx/vp9/common/vp9_frame_buffers.c', + 'libvpx/vp9/common/vp9_idct.c', + 'libvpx/vp9/common/vp9_loopfilter.c', + 'libvpx/vp9/common/vp9_mfqe.c', + 'libvpx/vp9/common/vp9_mvref_common.c', + 'libvpx/vp9/common/vp9_postproc.c', + 'libvpx/vp9/common/vp9_pred_common.c', + 'libvpx/vp9/common/vp9_quant_common.c', + 'libvpx/vp9/common/vp9_reconinter.c', + 'libvpx/vp9/common/vp9_reconintra.c', + 'libvpx/vp9/common/vp9_rtcd.c', + 'libvpx/vp9/common/vp9_scale.c', + 'libvpx/vp9/common/vp9_scan.c', + 'libvpx/vp9/common/vp9_seg_common.c', + 'libvpx/vp9/common/vp9_thread_common.c', + 'libvpx/vp9/common/vp9_tile_common.c', + 'libvpx/vp9/common/x86/vp9_idct_intrin_sse2.c', + 'libvpx/vp9/common/x86/vp9_mfqe_sse2.asm', + 'libvpx/vp9/decoder/vp9_decodeframe.c', + 'libvpx/vp9/decoder/vp9_decodemv.c', + 'libvpx/vp9/decoder/vp9_decoder.c', + 'libvpx/vp9/decoder/vp9_detokenize.c', + 'libvpx/vp9/decoder/vp9_dsubexp.c', + 'libvpx/vp9/decoder/vp9_job_queue.c', + 'libvpx/vp9/encoder/vp9_alt_ref_aq.c', + 'libvpx/vp9/encoder/vp9_aq_360.c', + 'libvpx/vp9/encoder/vp9_aq_complexity.c', + 'libvpx/vp9/encoder/vp9_aq_cyclicrefresh.c', + 'libvpx/vp9/encoder/vp9_aq_variance.c', + 'libvpx/vp9/encoder/vp9_bitstream.c', + 'libvpx/vp9/encoder/vp9_context_tree.c', + 'libvpx/vp9/encoder/vp9_cost.c', + 'libvpx/vp9/encoder/vp9_dct.c', + 'libvpx/vp9/encoder/vp9_encodeframe.c', + 'libvpx/vp9/encoder/vp9_encodemb.c', + 'libvpx/vp9/encoder/vp9_encodemv.c', + 'libvpx/vp9/encoder/vp9_encoder.c', + 'libvpx/vp9/encoder/vp9_ethread.c', + 'libvpx/vp9/encoder/vp9_ext_ratectrl.c', + 'libvpx/vp9/encoder/vp9_extend.c', + 'libvpx/vp9/encoder/vp9_firstpass.c', + 'libvpx/vp9/encoder/vp9_frame_scale.c', + 'libvpx/vp9/encoder/vp9_lookahead.c', + 'libvpx/vp9/encoder/vp9_mbgraph.c', + 'libvpx/vp9/encoder/vp9_mcomp.c', + 'libvpx/vp9/encoder/vp9_multi_thread.c', + 'libvpx/vp9/encoder/vp9_noise_estimate.c', + 'libvpx/vp9/encoder/vp9_picklpf.c', + 'libvpx/vp9/encoder/vp9_pickmode.c', + 'libvpx/vp9/encoder/vp9_quantize.c', + 'libvpx/vp9/encoder/vp9_ratectrl.c', + 'libvpx/vp9/encoder/vp9_rd.c', + 'libvpx/vp9/encoder/vp9_rdopt.c', + 'libvpx/vp9/encoder/vp9_resize.c', + 'libvpx/vp9/encoder/vp9_segmentation.c', + 'libvpx/vp9/encoder/vp9_skin_detection.c', + 'libvpx/vp9/encoder/vp9_speed_features.c', + 'libvpx/vp9/encoder/vp9_subexp.c', + 'libvpx/vp9/encoder/vp9_svc_layercontext.c', + 'libvpx/vp9/encoder/vp9_temporal_filter.c', + 'libvpx/vp9/encoder/vp9_tokenize.c', + 'libvpx/vp9/encoder/vp9_tpl_model.c', + 'libvpx/vp9/encoder/vp9_treewriter.c', + 'libvpx/vp9/encoder/x86/temporal_filter_sse4.c', + 'libvpx/vp9/encoder/x86/vp9_dct_intrin_sse2.c', + 'libvpx/vp9/encoder/x86/vp9_dct_sse2.asm', + 'libvpx/vp9/encoder/x86/vp9_error_avx2.c', + 'libvpx/vp9/encoder/x86/vp9_error_sse2.asm', + 'libvpx/vp9/encoder/x86/vp9_frame_scale_ssse3.c', + 'libvpx/vp9/encoder/x86/vp9_quantize_avx2.c', + 'libvpx/vp9/encoder/x86/vp9_quantize_sse2.c', + 'libvpx/vp9/encoder/x86/vp9_quantize_ssse3.c', + 'libvpx/vp9/vp9_cx_iface.c', + 'libvpx/vp9/vp9_dx_iface.c', + 'libvpx/vp9/vp9_iface_common.c', + 'libvpx/vpx/src/vpx_codec.c', + 'libvpx/vpx/src/vpx_decoder.c', + 'libvpx/vpx/src/vpx_encoder.c', + 'libvpx/vpx/src/vpx_image.c', + 'libvpx/vpx_dsp/add_noise.c', + 'libvpx/vpx_dsp/avg.c', + 'libvpx/vpx_dsp/bitreader.c', + 'libvpx/vpx_dsp/bitreader_buffer.c', + 'libvpx/vpx_dsp/bitwriter.c', + 'libvpx/vpx_dsp/bitwriter_buffer.c', + 'libvpx/vpx_dsp/deblock.c', + 'libvpx/vpx_dsp/fwd_txfm.c', + 'libvpx/vpx_dsp/intrapred.c', + 'libvpx/vpx_dsp/inv_txfm.c', + 'libvpx/vpx_dsp/loopfilter.c', + 'libvpx/vpx_dsp/prob.c', + 'libvpx/vpx_dsp/psnr.c', + 'libvpx/vpx_dsp/quantize.c', + 'libvpx/vpx_dsp/sad.c', + 'libvpx/vpx_dsp/skin_detection.c', + 'libvpx/vpx_dsp/sse.c', + 'libvpx/vpx_dsp/subtract.c', + 'libvpx/vpx_dsp/sum_squares.c', + 'libvpx/vpx_dsp/variance.c', + 'libvpx/vpx_dsp/vpx_convolve.c', + 'libvpx/vpx_dsp/vpx_dsp_rtcd.c', + 'libvpx/vpx_dsp/x86/add_noise_sse2.asm', + 'libvpx/vpx_dsp/x86/avg_intrin_avx2.c', + 'libvpx/vpx_dsp/x86/avg_intrin_sse2.c', + 'libvpx/vpx_dsp/x86/avg_pred_avx2.c', + 'libvpx/vpx_dsp/x86/avg_pred_sse2.c', + 'libvpx/vpx_dsp/x86/avg_ssse3_x86_64.asm', + 'libvpx/vpx_dsp/x86/deblock_sse2.asm', + 'libvpx/vpx_dsp/x86/fwd_txfm_avx2.c', + 'libvpx/vpx_dsp/x86/fwd_txfm_sse2.c', + 'libvpx/vpx_dsp/x86/fwd_txfm_ssse3_x86_64.asm', + 'libvpx/vpx_dsp/x86/intrapred_sse2.asm', + 'libvpx/vpx_dsp/x86/intrapred_ssse3.asm', + 'libvpx/vpx_dsp/x86/inv_txfm_avx2.c', + 'libvpx/vpx_dsp/x86/inv_txfm_sse2.c', + 'libvpx/vpx_dsp/x86/inv_txfm_ssse3.c', + 'libvpx/vpx_dsp/x86/inv_wht_sse2.asm', + 'libvpx/vpx_dsp/x86/loopfilter_avx2.c', + 'libvpx/vpx_dsp/x86/loopfilter_intrin_sse2.c', + 'libvpx/vpx_dsp/x86/post_proc_sse2.c', + 'libvpx/vpx_dsp/x86/quantize_avx.c', + 'libvpx/vpx_dsp/x86/quantize_avx2.c', + 'libvpx/vpx_dsp/x86/quantize_sse2.c', + 'libvpx/vpx_dsp/x86/quantize_ssse3.c', + 'libvpx/vpx_dsp/x86/sad4d_avx2.c', + 'libvpx/vpx_dsp/x86/sad4d_sse2.asm', + 'libvpx/vpx_dsp/x86/sad_avx2.c', + 'libvpx/vpx_dsp/x86/sad_sse2.asm', + 'libvpx/vpx_dsp/x86/sse_avx2.c', + 'libvpx/vpx_dsp/x86/sse_sse4.c', + 'libvpx/vpx_dsp/x86/ssim_opt_x86_64.asm', + 'libvpx/vpx_dsp/x86/subpel_variance_sse2.asm', + 'libvpx/vpx_dsp/x86/subtract_avx2.c', + 'libvpx/vpx_dsp/x86/subtract_sse2.asm', + 'libvpx/vpx_dsp/x86/sum_squares_sse2.c', + 'libvpx/vpx_dsp/x86/variance_avx2.c', + 'libvpx/vpx_dsp/x86/variance_sse2.c', + 'libvpx/vpx_dsp/x86/vpx_convolve_copy_sse2.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_4t_intrin_sse2.c', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_intrin_avx2.c', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_intrin_ssse3.c', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_sse2.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_ssse3.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_bilinear_sse2.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_bilinear_ssse3.asm', + 'libvpx/vpx_mem/vpx_mem.c', + 'libvpx/vpx_ports/emms_mmx.asm', + 'libvpx/vpx_ports/float_control_word.asm', + 'libvpx/vpx_scale/generic/gen_scalers.c', + 'libvpx/vpx_scale/generic/vpx_scale.c', + 'libvpx/vpx_scale/generic/yv12config.c', + 'libvpx/vpx_scale/generic/yv12extend.c', + 'libvpx/vpx_scale/vpx_scale_rtcd.c', + 'libvpx/vpx_util/vpx_thread.c', + 'libvpx/vpx_util/vpx_write_yuv_frame.c', +], + 'WIN_X64_EXPORTS': [ + 'libvpx/vpx/vp8.h', + 'libvpx/vpx/vp8cx.h', + 'libvpx/vpx/vp8dx.h', + 'libvpx/vpx/vpx_codec.h', + 'libvpx/vpx/vpx_decoder.h', + 'libvpx/vpx/vpx_encoder.h', + 'libvpx/vpx/vpx_ext_ratectrl.h', + 'libvpx/vpx/vpx_frame_buffer.h', + 'libvpx/vpx/vpx_image.h', + 'libvpx/vpx/vpx_integer.h', + 'libvpx/vpx/vpx_tpl.h', + 'libvpx/vpx_mem/include/vpx_mem_intrnl.h', + 'libvpx/vpx_mem/vpx_mem.h', + 'libvpx/vpx_ports/bitops.h', + 'libvpx/vpx_ports/compiler_attributes.h', + 'libvpx/vpx_ports/mem.h', + 'libvpx/vpx_ports/static_assert.h', + 'libvpx/vpx_ports/system_state.h', + 'libvpx/vpx_ports/vpx_timer.h', + 'libvpx/vpx_ports/x86.h', + 'libvpx/vpx_scale/vpx_scale.h', + 'libvpx/vpx_scale/yv12config.h', +], + 'WIN_X64_SOURCES': [ + 'libvpx/../config/win/x64/vpx_config.c', + 'libvpx/vp8/common/alloccommon.c', + 'libvpx/vp8/common/blockd.c', + 'libvpx/vp8/common/dequantize.c', + 'libvpx/vp8/common/entropy.c', + 'libvpx/vp8/common/entropymode.c', + 'libvpx/vp8/common/entropymv.c', + 'libvpx/vp8/common/extend.c', + 'libvpx/vp8/common/filter.c', + 'libvpx/vp8/common/findnearmv.c', + 'libvpx/vp8/common/generic/systemdependent.c', + 'libvpx/vp8/common/idct_blk.c', + 'libvpx/vp8/common/idctllm.c', + 'libvpx/vp8/common/loopfilter_filters.c', + 'libvpx/vp8/common/mbpitch.c', + 'libvpx/vp8/common/mfqe.c', + 'libvpx/vp8/common/modecont.c', + 'libvpx/vp8/common/postproc.c', + 'libvpx/vp8/common/quant_common.c', + 'libvpx/vp8/common/reconinter.c', + 'libvpx/vp8/common/reconintra.c', + 'libvpx/vp8/common/reconintra4x4.c', + 'libvpx/vp8/common/rtcd.c', + 'libvpx/vp8/common/setupintrarecon.c', + 'libvpx/vp8/common/swapyv12buffer.c', + 'libvpx/vp8/common/treecoder.c', + 'libvpx/vp8/common/vp8_loopfilter.c', + 'libvpx/vp8/common/vp8_skin_detection.c', + 'libvpx/vp8/common/x86/bilinear_filter_sse2.c', + 'libvpx/vp8/common/x86/dequantize_mmx.asm', + 'libvpx/vp8/common/x86/idct_blk_mmx.c', + 'libvpx/vp8/common/x86/idct_blk_sse2.c', + 'libvpx/vp8/common/x86/idctllm_mmx.asm', + 'libvpx/vp8/common/x86/idctllm_sse2.asm', + 'libvpx/vp8/common/x86/iwalsh_sse2.asm', + 'libvpx/vp8/common/x86/loopfilter_block_sse2_x86_64.asm', + 'libvpx/vp8/common/x86/loopfilter_sse2.asm', + 'libvpx/vp8/common/x86/loopfilter_x86.c', + 'libvpx/vp8/common/x86/mfqe_sse2.asm', + 'libvpx/vp8/common/x86/recon_mmx.asm', + 'libvpx/vp8/common/x86/recon_sse2.asm', + 'libvpx/vp8/common/x86/subpixel_mmx.asm', + 'libvpx/vp8/common/x86/subpixel_sse2.asm', + 'libvpx/vp8/common/x86/subpixel_ssse3.asm', + 'libvpx/vp8/common/x86/vp8_asm_stubs.c', + 'libvpx/vp8/decoder/dboolhuff.c', + 'libvpx/vp8/decoder/decodeframe.c', + 'libvpx/vp8/decoder/decodemv.c', + 'libvpx/vp8/decoder/detokenize.c', + 'libvpx/vp8/decoder/onyxd_if.c', + 'libvpx/vp8/decoder/threading.c', + 'libvpx/vp8/encoder/bitstream.c', + 'libvpx/vp8/encoder/boolhuff.c', + 'libvpx/vp8/encoder/copy_c.c', + 'libvpx/vp8/encoder/dct.c', + 'libvpx/vp8/encoder/denoising.c', + 'libvpx/vp8/encoder/encodeframe.c', + 'libvpx/vp8/encoder/encodeintra.c', + 'libvpx/vp8/encoder/encodemb.c', + 'libvpx/vp8/encoder/encodemv.c', + 'libvpx/vp8/encoder/ethreading.c', + 'libvpx/vp8/encoder/firstpass.c', + 'libvpx/vp8/encoder/lookahead.c', + 'libvpx/vp8/encoder/mcomp.c', + 'libvpx/vp8/encoder/modecosts.c', + 'libvpx/vp8/encoder/mr_dissim.c', + 'libvpx/vp8/encoder/onyx_if.c', + 'libvpx/vp8/encoder/pickinter.c', + 'libvpx/vp8/encoder/picklpf.c', + 'libvpx/vp8/encoder/ratectrl.c', + 'libvpx/vp8/encoder/rdopt.c', + 'libvpx/vp8/encoder/segmentation.c', + 'libvpx/vp8/encoder/temporal_filter.c', + 'libvpx/vp8/encoder/tokenize.c', + 'libvpx/vp8/encoder/treewriter.c', + 'libvpx/vp8/encoder/vp8_quantize.c', + 'libvpx/vp8/encoder/x86/block_error_sse2.asm', + 'libvpx/vp8/encoder/x86/copy_sse2.asm', + 'libvpx/vp8/encoder/x86/copy_sse3.asm', + 'libvpx/vp8/encoder/x86/dct_sse2.asm', + 'libvpx/vp8/encoder/x86/denoising_sse2.c', + 'libvpx/vp8/encoder/x86/fwalsh_sse2.asm', + 'libvpx/vp8/encoder/x86/quantize_sse4.c', + 'libvpx/vp8/encoder/x86/temporal_filter_apply_sse2.asm', + 'libvpx/vp8/encoder/x86/vp8_enc_stubs_sse2.c', + 'libvpx/vp8/encoder/x86/vp8_quantize_sse2.c', + 'libvpx/vp8/encoder/x86/vp8_quantize_ssse3.c', + 'libvpx/vp8/vp8_cx_iface.c', + 'libvpx/vp8/vp8_dx_iface.c', + 'libvpx/vp9/common/vp9_alloccommon.c', + 'libvpx/vp9/common/vp9_blockd.c', + 'libvpx/vp9/common/vp9_common_data.c', + 'libvpx/vp9/common/vp9_entropy.c', + 'libvpx/vp9/common/vp9_entropymode.c', + 'libvpx/vp9/common/vp9_entropymv.c', + 'libvpx/vp9/common/vp9_filter.c', + 'libvpx/vp9/common/vp9_frame_buffers.c', + 'libvpx/vp9/common/vp9_idct.c', + 'libvpx/vp9/common/vp9_loopfilter.c', + 'libvpx/vp9/common/vp9_mfqe.c', + 'libvpx/vp9/common/vp9_mvref_common.c', + 'libvpx/vp9/common/vp9_postproc.c', + 'libvpx/vp9/common/vp9_pred_common.c', + 'libvpx/vp9/common/vp9_quant_common.c', + 'libvpx/vp9/common/vp9_reconinter.c', + 'libvpx/vp9/common/vp9_reconintra.c', + 'libvpx/vp9/common/vp9_rtcd.c', + 'libvpx/vp9/common/vp9_scale.c', + 'libvpx/vp9/common/vp9_scan.c', + 'libvpx/vp9/common/vp9_seg_common.c', + 'libvpx/vp9/common/vp9_thread_common.c', + 'libvpx/vp9/common/vp9_tile_common.c', + 'libvpx/vp9/common/x86/vp9_idct_intrin_sse2.c', + 'libvpx/vp9/common/x86/vp9_mfqe_sse2.asm', + 'libvpx/vp9/decoder/vp9_decodeframe.c', + 'libvpx/vp9/decoder/vp9_decodemv.c', + 'libvpx/vp9/decoder/vp9_decoder.c', + 'libvpx/vp9/decoder/vp9_detokenize.c', + 'libvpx/vp9/decoder/vp9_dsubexp.c', + 'libvpx/vp9/decoder/vp9_job_queue.c', + 'libvpx/vp9/encoder/vp9_alt_ref_aq.c', + 'libvpx/vp9/encoder/vp9_aq_360.c', + 'libvpx/vp9/encoder/vp9_aq_complexity.c', + 'libvpx/vp9/encoder/vp9_aq_cyclicrefresh.c', + 'libvpx/vp9/encoder/vp9_aq_variance.c', + 'libvpx/vp9/encoder/vp9_bitstream.c', + 'libvpx/vp9/encoder/vp9_context_tree.c', + 'libvpx/vp9/encoder/vp9_cost.c', + 'libvpx/vp9/encoder/vp9_dct.c', + 'libvpx/vp9/encoder/vp9_encodeframe.c', + 'libvpx/vp9/encoder/vp9_encodemb.c', + 'libvpx/vp9/encoder/vp9_encodemv.c', + 'libvpx/vp9/encoder/vp9_encoder.c', + 'libvpx/vp9/encoder/vp9_ethread.c', + 'libvpx/vp9/encoder/vp9_ext_ratectrl.c', + 'libvpx/vp9/encoder/vp9_extend.c', + 'libvpx/vp9/encoder/vp9_firstpass.c', + 'libvpx/vp9/encoder/vp9_frame_scale.c', + 'libvpx/vp9/encoder/vp9_lookahead.c', + 'libvpx/vp9/encoder/vp9_mbgraph.c', + 'libvpx/vp9/encoder/vp9_mcomp.c', + 'libvpx/vp9/encoder/vp9_multi_thread.c', + 'libvpx/vp9/encoder/vp9_noise_estimate.c', + 'libvpx/vp9/encoder/vp9_picklpf.c', + 'libvpx/vp9/encoder/vp9_pickmode.c', + 'libvpx/vp9/encoder/vp9_quantize.c', + 'libvpx/vp9/encoder/vp9_ratectrl.c', + 'libvpx/vp9/encoder/vp9_rd.c', + 'libvpx/vp9/encoder/vp9_rdopt.c', + 'libvpx/vp9/encoder/vp9_resize.c', + 'libvpx/vp9/encoder/vp9_segmentation.c', + 'libvpx/vp9/encoder/vp9_skin_detection.c', + 'libvpx/vp9/encoder/vp9_speed_features.c', + 'libvpx/vp9/encoder/vp9_subexp.c', + 'libvpx/vp9/encoder/vp9_svc_layercontext.c', + 'libvpx/vp9/encoder/vp9_temporal_filter.c', + 'libvpx/vp9/encoder/vp9_tokenize.c', + 'libvpx/vp9/encoder/vp9_tpl_model.c', + 'libvpx/vp9/encoder/vp9_treewriter.c', + 'libvpx/vp9/encoder/x86/temporal_filter_sse4.c', + 'libvpx/vp9/encoder/x86/vp9_dct_intrin_sse2.c', + 'libvpx/vp9/encoder/x86/vp9_dct_sse2.asm', + 'libvpx/vp9/encoder/x86/vp9_error_avx2.c', + 'libvpx/vp9/encoder/x86/vp9_error_sse2.asm', + 'libvpx/vp9/encoder/x86/vp9_frame_scale_ssse3.c', + 'libvpx/vp9/encoder/x86/vp9_quantize_avx2.c', + 'libvpx/vp9/encoder/x86/vp9_quantize_sse2.c', + 'libvpx/vp9/encoder/x86/vp9_quantize_ssse3.c', + 'libvpx/vp9/vp9_cx_iface.c', + 'libvpx/vp9/vp9_dx_iface.c', + 'libvpx/vp9/vp9_iface_common.c', + 'libvpx/vpx/src/vpx_codec.c', + 'libvpx/vpx/src/vpx_decoder.c', + 'libvpx/vpx/src/vpx_encoder.c', + 'libvpx/vpx/src/vpx_image.c', + 'libvpx/vpx_dsp/add_noise.c', + 'libvpx/vpx_dsp/avg.c', + 'libvpx/vpx_dsp/bitreader.c', + 'libvpx/vpx_dsp/bitreader_buffer.c', + 'libvpx/vpx_dsp/bitwriter.c', + 'libvpx/vpx_dsp/bitwriter_buffer.c', + 'libvpx/vpx_dsp/deblock.c', + 'libvpx/vpx_dsp/fwd_txfm.c', + 'libvpx/vpx_dsp/intrapred.c', + 'libvpx/vpx_dsp/inv_txfm.c', + 'libvpx/vpx_dsp/loopfilter.c', + 'libvpx/vpx_dsp/prob.c', + 'libvpx/vpx_dsp/psnr.c', + 'libvpx/vpx_dsp/quantize.c', + 'libvpx/vpx_dsp/sad.c', + 'libvpx/vpx_dsp/skin_detection.c', + 'libvpx/vpx_dsp/sse.c', + 'libvpx/vpx_dsp/subtract.c', + 'libvpx/vpx_dsp/sum_squares.c', + 'libvpx/vpx_dsp/variance.c', + 'libvpx/vpx_dsp/vpx_convolve.c', + 'libvpx/vpx_dsp/vpx_dsp_rtcd.c', + 'libvpx/vpx_dsp/x86/add_noise_sse2.asm', + 'libvpx/vpx_dsp/x86/avg_intrin_avx2.c', + 'libvpx/vpx_dsp/x86/avg_intrin_sse2.c', + 'libvpx/vpx_dsp/x86/avg_pred_avx2.c', + 'libvpx/vpx_dsp/x86/avg_pred_sse2.c', + 'libvpx/vpx_dsp/x86/avg_ssse3_x86_64.asm', + 'libvpx/vpx_dsp/x86/deblock_sse2.asm', + 'libvpx/vpx_dsp/x86/fwd_txfm_avx2.c', + 'libvpx/vpx_dsp/x86/fwd_txfm_sse2.c', + 'libvpx/vpx_dsp/x86/fwd_txfm_ssse3_x86_64.asm', + 'libvpx/vpx_dsp/x86/intrapred_sse2.asm', + 'libvpx/vpx_dsp/x86/intrapred_ssse3.asm', + 'libvpx/vpx_dsp/x86/inv_txfm_avx2.c', + 'libvpx/vpx_dsp/x86/inv_txfm_sse2.c', + 'libvpx/vpx_dsp/x86/inv_txfm_ssse3.c', + 'libvpx/vpx_dsp/x86/inv_wht_sse2.asm', + 'libvpx/vpx_dsp/x86/loopfilter_avx2.c', + 'libvpx/vpx_dsp/x86/loopfilter_intrin_sse2.c', + 'libvpx/vpx_dsp/x86/post_proc_sse2.c', + 'libvpx/vpx_dsp/x86/quantize_avx.c', + 'libvpx/vpx_dsp/x86/quantize_avx2.c', + 'libvpx/vpx_dsp/x86/quantize_sse2.c', + 'libvpx/vpx_dsp/x86/quantize_ssse3.c', + 'libvpx/vpx_dsp/x86/sad4d_avx2.c', + 'libvpx/vpx_dsp/x86/sad4d_sse2.asm', + 'libvpx/vpx_dsp/x86/sad_avx2.c', + 'libvpx/vpx_dsp/x86/sad_sse2.asm', + 'libvpx/vpx_dsp/x86/sse_avx2.c', + 'libvpx/vpx_dsp/x86/sse_sse4.c', + 'libvpx/vpx_dsp/x86/ssim_opt_x86_64.asm', + 'libvpx/vpx_dsp/x86/subpel_variance_sse2.asm', + 'libvpx/vpx_dsp/x86/subtract_avx2.c', + 'libvpx/vpx_dsp/x86/subtract_sse2.asm', + 'libvpx/vpx_dsp/x86/sum_squares_sse2.c', + 'libvpx/vpx_dsp/x86/variance_avx2.c', + 'libvpx/vpx_dsp/x86/variance_sse2.c', + 'libvpx/vpx_dsp/x86/vpx_convolve_copy_sse2.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_4t_intrin_sse2.c', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_intrin_avx2.c', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_intrin_ssse3.c', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_sse2.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_ssse3.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_bilinear_sse2.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_bilinear_ssse3.asm', + 'libvpx/vpx_mem/vpx_mem.c', + 'libvpx/vpx_ports/emms_mmx.asm', + 'libvpx/vpx_ports/float_control_word.asm', + 'libvpx/vpx_scale/generic/gen_scalers.c', + 'libvpx/vpx_scale/generic/vpx_scale.c', + 'libvpx/vpx_scale/generic/yv12config.c', + 'libvpx/vpx_scale/generic/yv12extend.c', + 'libvpx/vpx_scale/vpx_scale_rtcd.c', + 'libvpx/vpx_util/vpx_thread.c', + 'libvpx/vpx_util/vpx_write_yuv_frame.c', +], + 'LINUX_IA32_EXPORTS': [ + 'libvpx/vpx/vp8.h', + 'libvpx/vpx/vp8cx.h', + 'libvpx/vpx/vp8dx.h', + 'libvpx/vpx/vpx_codec.h', + 'libvpx/vpx/vpx_decoder.h', + 'libvpx/vpx/vpx_encoder.h', + 'libvpx/vpx/vpx_ext_ratectrl.h', + 'libvpx/vpx/vpx_frame_buffer.h', + 'libvpx/vpx/vpx_image.h', + 'libvpx/vpx/vpx_integer.h', + 'libvpx/vpx/vpx_tpl.h', + 'libvpx/vpx_mem/include/vpx_mem_intrnl.h', + 'libvpx/vpx_mem/vpx_mem.h', + 'libvpx/vpx_ports/bitops.h', + 'libvpx/vpx_ports/compiler_attributes.h', + 'libvpx/vpx_ports/mem.h', + 'libvpx/vpx_ports/static_assert.h', + 'libvpx/vpx_ports/system_state.h', + 'libvpx/vpx_ports/vpx_timer.h', + 'libvpx/vpx_ports/x86.h', + 'libvpx/vpx_scale/vpx_scale.h', + 'libvpx/vpx_scale/yv12config.h', +], + 'LINUX_IA32_SOURCES': [ + 'libvpx/../config/linux/ia32/vpx_config.c', 'libvpx/vp8/common/alloccommon.c', 'libvpx/vp8/common/blockd.c', 'libvpx/vp8/common/dequantize.c', @@ -476,7 +1031,6 @@ files = { 'libvpx/vpx/src/vpx_decoder.c', 'libvpx/vpx/src/vpx_encoder.c', 'libvpx/vpx/src/vpx_image.c', - 'libvpx/vpx/src/vpx_tpl.c', 'libvpx/vpx_dsp/add_noise.c', 'libvpx/vpx_dsp/avg.c', 'libvpx/vpx_dsp/bitreader.c', @@ -550,7 +1104,551 @@ files = { 'libvpx/vpx_util/vpx_thread.c', 'libvpx/vpx_util/vpx_write_yuv_frame.c', ], - 'ARM_EXPORTS': [ + 'MAC_IA32_EXPORTS': [ + 'libvpx/vpx/vp8.h', + 'libvpx/vpx/vp8cx.h', + 'libvpx/vpx/vp8dx.h', + 'libvpx/vpx/vpx_codec.h', + 'libvpx/vpx/vpx_decoder.h', + 'libvpx/vpx/vpx_encoder.h', + 'libvpx/vpx/vpx_ext_ratectrl.h', + 'libvpx/vpx/vpx_frame_buffer.h', + 'libvpx/vpx/vpx_image.h', + 'libvpx/vpx/vpx_integer.h', + 'libvpx/vpx/vpx_tpl.h', + 'libvpx/vpx_mem/include/vpx_mem_intrnl.h', + 'libvpx/vpx_mem/vpx_mem.h', + 'libvpx/vpx_ports/bitops.h', + 'libvpx/vpx_ports/compiler_attributes.h', + 'libvpx/vpx_ports/mem.h', + 'libvpx/vpx_ports/static_assert.h', + 'libvpx/vpx_ports/system_state.h', + 'libvpx/vpx_ports/vpx_timer.h', + 'libvpx/vpx_ports/x86.h', + 'libvpx/vpx_scale/vpx_scale.h', + 'libvpx/vpx_scale/yv12config.h', +], + 'MAC_IA32_SOURCES': [ + 'libvpx/../config/mac/ia32/vpx_config.c', + 'libvpx/vp8/common/alloccommon.c', + 'libvpx/vp8/common/blockd.c', + 'libvpx/vp8/common/dequantize.c', + 'libvpx/vp8/common/entropy.c', + 'libvpx/vp8/common/entropymode.c', + 'libvpx/vp8/common/entropymv.c', + 'libvpx/vp8/common/extend.c', + 'libvpx/vp8/common/filter.c', + 'libvpx/vp8/common/findnearmv.c', + 'libvpx/vp8/common/generic/systemdependent.c', + 'libvpx/vp8/common/idct_blk.c', + 'libvpx/vp8/common/idctllm.c', + 'libvpx/vp8/common/loopfilter_filters.c', + 'libvpx/vp8/common/mbpitch.c', + 'libvpx/vp8/common/mfqe.c', + 'libvpx/vp8/common/modecont.c', + 'libvpx/vp8/common/postproc.c', + 'libvpx/vp8/common/quant_common.c', + 'libvpx/vp8/common/reconinter.c', + 'libvpx/vp8/common/reconintra.c', + 'libvpx/vp8/common/reconintra4x4.c', + 'libvpx/vp8/common/rtcd.c', + 'libvpx/vp8/common/setupintrarecon.c', + 'libvpx/vp8/common/swapyv12buffer.c', + 'libvpx/vp8/common/treecoder.c', + 'libvpx/vp8/common/vp8_loopfilter.c', + 'libvpx/vp8/common/vp8_skin_detection.c', + 'libvpx/vp8/common/x86/bilinear_filter_sse2.c', + 'libvpx/vp8/common/x86/dequantize_mmx.asm', + 'libvpx/vp8/common/x86/idct_blk_mmx.c', + 'libvpx/vp8/common/x86/idct_blk_sse2.c', + 'libvpx/vp8/common/x86/idctllm_mmx.asm', + 'libvpx/vp8/common/x86/idctllm_sse2.asm', + 'libvpx/vp8/common/x86/iwalsh_sse2.asm', + 'libvpx/vp8/common/x86/loopfilter_sse2.asm', + 'libvpx/vp8/common/x86/loopfilter_x86.c', + 'libvpx/vp8/common/x86/mfqe_sse2.asm', + 'libvpx/vp8/common/x86/recon_mmx.asm', + 'libvpx/vp8/common/x86/recon_sse2.asm', + 'libvpx/vp8/common/x86/subpixel_mmx.asm', + 'libvpx/vp8/common/x86/subpixel_sse2.asm', + 'libvpx/vp8/common/x86/subpixel_ssse3.asm', + 'libvpx/vp8/common/x86/vp8_asm_stubs.c', + 'libvpx/vp8/decoder/dboolhuff.c', + 'libvpx/vp8/decoder/decodeframe.c', + 'libvpx/vp8/decoder/decodemv.c', + 'libvpx/vp8/decoder/detokenize.c', + 'libvpx/vp8/decoder/onyxd_if.c', + 'libvpx/vp8/decoder/threading.c', + 'libvpx/vp8/encoder/bitstream.c', + 'libvpx/vp8/encoder/boolhuff.c', + 'libvpx/vp8/encoder/copy_c.c', + 'libvpx/vp8/encoder/dct.c', + 'libvpx/vp8/encoder/denoising.c', + 'libvpx/vp8/encoder/encodeframe.c', + 'libvpx/vp8/encoder/encodeintra.c', + 'libvpx/vp8/encoder/encodemb.c', + 'libvpx/vp8/encoder/encodemv.c', + 'libvpx/vp8/encoder/ethreading.c', + 'libvpx/vp8/encoder/firstpass.c', + 'libvpx/vp8/encoder/lookahead.c', + 'libvpx/vp8/encoder/mcomp.c', + 'libvpx/vp8/encoder/modecosts.c', + 'libvpx/vp8/encoder/mr_dissim.c', + 'libvpx/vp8/encoder/onyx_if.c', + 'libvpx/vp8/encoder/pickinter.c', + 'libvpx/vp8/encoder/picklpf.c', + 'libvpx/vp8/encoder/ratectrl.c', + 'libvpx/vp8/encoder/rdopt.c', + 'libvpx/vp8/encoder/segmentation.c', + 'libvpx/vp8/encoder/temporal_filter.c', + 'libvpx/vp8/encoder/tokenize.c', + 'libvpx/vp8/encoder/treewriter.c', + 'libvpx/vp8/encoder/vp8_quantize.c', + 'libvpx/vp8/encoder/x86/block_error_sse2.asm', + 'libvpx/vp8/encoder/x86/copy_sse2.asm', + 'libvpx/vp8/encoder/x86/copy_sse3.asm', + 'libvpx/vp8/encoder/x86/dct_sse2.asm', + 'libvpx/vp8/encoder/x86/denoising_sse2.c', + 'libvpx/vp8/encoder/x86/fwalsh_sse2.asm', + 'libvpx/vp8/encoder/x86/quantize_sse4.c', + 'libvpx/vp8/encoder/x86/temporal_filter_apply_sse2.asm', + 'libvpx/vp8/encoder/x86/vp8_enc_stubs_sse2.c', + 'libvpx/vp8/encoder/x86/vp8_quantize_sse2.c', + 'libvpx/vp8/encoder/x86/vp8_quantize_ssse3.c', + 'libvpx/vp8/vp8_cx_iface.c', + 'libvpx/vp8/vp8_dx_iface.c', + 'libvpx/vp9/common/vp9_alloccommon.c', + 'libvpx/vp9/common/vp9_blockd.c', + 'libvpx/vp9/common/vp9_common_data.c', + 'libvpx/vp9/common/vp9_entropy.c', + 'libvpx/vp9/common/vp9_entropymode.c', + 'libvpx/vp9/common/vp9_entropymv.c', + 'libvpx/vp9/common/vp9_filter.c', + 'libvpx/vp9/common/vp9_frame_buffers.c', + 'libvpx/vp9/common/vp9_idct.c', + 'libvpx/vp9/common/vp9_loopfilter.c', + 'libvpx/vp9/common/vp9_mfqe.c', + 'libvpx/vp9/common/vp9_mvref_common.c', + 'libvpx/vp9/common/vp9_postproc.c', + 'libvpx/vp9/common/vp9_pred_common.c', + 'libvpx/vp9/common/vp9_quant_common.c', + 'libvpx/vp9/common/vp9_reconinter.c', + 'libvpx/vp9/common/vp9_reconintra.c', + 'libvpx/vp9/common/vp9_rtcd.c', + 'libvpx/vp9/common/vp9_scale.c', + 'libvpx/vp9/common/vp9_scan.c', + 'libvpx/vp9/common/vp9_seg_common.c', + 'libvpx/vp9/common/vp9_thread_common.c', + 'libvpx/vp9/common/vp9_tile_common.c', + 'libvpx/vp9/common/x86/vp9_idct_intrin_sse2.c', + 'libvpx/vp9/common/x86/vp9_mfqe_sse2.asm', + 'libvpx/vp9/decoder/vp9_decodeframe.c', + 'libvpx/vp9/decoder/vp9_decodemv.c', + 'libvpx/vp9/decoder/vp9_decoder.c', + 'libvpx/vp9/decoder/vp9_detokenize.c', + 'libvpx/vp9/decoder/vp9_dsubexp.c', + 'libvpx/vp9/decoder/vp9_job_queue.c', + 'libvpx/vp9/encoder/vp9_alt_ref_aq.c', + 'libvpx/vp9/encoder/vp9_aq_360.c', + 'libvpx/vp9/encoder/vp9_aq_complexity.c', + 'libvpx/vp9/encoder/vp9_aq_cyclicrefresh.c', + 'libvpx/vp9/encoder/vp9_aq_variance.c', + 'libvpx/vp9/encoder/vp9_bitstream.c', + 'libvpx/vp9/encoder/vp9_context_tree.c', + 'libvpx/vp9/encoder/vp9_cost.c', + 'libvpx/vp9/encoder/vp9_dct.c', + 'libvpx/vp9/encoder/vp9_encodeframe.c', + 'libvpx/vp9/encoder/vp9_encodemb.c', + 'libvpx/vp9/encoder/vp9_encodemv.c', + 'libvpx/vp9/encoder/vp9_encoder.c', + 'libvpx/vp9/encoder/vp9_ethread.c', + 'libvpx/vp9/encoder/vp9_ext_ratectrl.c', + 'libvpx/vp9/encoder/vp9_extend.c', + 'libvpx/vp9/encoder/vp9_firstpass.c', + 'libvpx/vp9/encoder/vp9_frame_scale.c', + 'libvpx/vp9/encoder/vp9_lookahead.c', + 'libvpx/vp9/encoder/vp9_mbgraph.c', + 'libvpx/vp9/encoder/vp9_mcomp.c', + 'libvpx/vp9/encoder/vp9_multi_thread.c', + 'libvpx/vp9/encoder/vp9_noise_estimate.c', + 'libvpx/vp9/encoder/vp9_picklpf.c', + 'libvpx/vp9/encoder/vp9_pickmode.c', + 'libvpx/vp9/encoder/vp9_quantize.c', + 'libvpx/vp9/encoder/vp9_ratectrl.c', + 'libvpx/vp9/encoder/vp9_rd.c', + 'libvpx/vp9/encoder/vp9_rdopt.c', + 'libvpx/vp9/encoder/vp9_resize.c', + 'libvpx/vp9/encoder/vp9_segmentation.c', + 'libvpx/vp9/encoder/vp9_skin_detection.c', + 'libvpx/vp9/encoder/vp9_speed_features.c', + 'libvpx/vp9/encoder/vp9_subexp.c', + 'libvpx/vp9/encoder/vp9_svc_layercontext.c', + 'libvpx/vp9/encoder/vp9_temporal_filter.c', + 'libvpx/vp9/encoder/vp9_tokenize.c', + 'libvpx/vp9/encoder/vp9_tpl_model.c', + 'libvpx/vp9/encoder/vp9_treewriter.c', + 'libvpx/vp9/encoder/x86/temporal_filter_sse4.c', + 'libvpx/vp9/encoder/x86/vp9_dct_intrin_sse2.c', + 'libvpx/vp9/encoder/x86/vp9_dct_sse2.asm', + 'libvpx/vp9/encoder/x86/vp9_error_avx2.c', + 'libvpx/vp9/encoder/x86/vp9_error_sse2.asm', + 'libvpx/vp9/encoder/x86/vp9_frame_scale_ssse3.c', + 'libvpx/vp9/encoder/x86/vp9_quantize_avx2.c', + 'libvpx/vp9/encoder/x86/vp9_quantize_sse2.c', + 'libvpx/vp9/encoder/x86/vp9_quantize_ssse3.c', + 'libvpx/vp9/vp9_cx_iface.c', + 'libvpx/vp9/vp9_dx_iface.c', + 'libvpx/vp9/vp9_iface_common.c', + 'libvpx/vpx/src/vpx_codec.c', + 'libvpx/vpx/src/vpx_decoder.c', + 'libvpx/vpx/src/vpx_encoder.c', + 'libvpx/vpx/src/vpx_image.c', + 'libvpx/vpx_dsp/add_noise.c', + 'libvpx/vpx_dsp/avg.c', + 'libvpx/vpx_dsp/bitreader.c', + 'libvpx/vpx_dsp/bitreader_buffer.c', + 'libvpx/vpx_dsp/bitwriter.c', + 'libvpx/vpx_dsp/bitwriter_buffer.c', + 'libvpx/vpx_dsp/deblock.c', + 'libvpx/vpx_dsp/fwd_txfm.c', + 'libvpx/vpx_dsp/intrapred.c', + 'libvpx/vpx_dsp/inv_txfm.c', + 'libvpx/vpx_dsp/loopfilter.c', + 'libvpx/vpx_dsp/prob.c', + 'libvpx/vpx_dsp/psnr.c', + 'libvpx/vpx_dsp/quantize.c', + 'libvpx/vpx_dsp/sad.c', + 'libvpx/vpx_dsp/skin_detection.c', + 'libvpx/vpx_dsp/sse.c', + 'libvpx/vpx_dsp/subtract.c', + 'libvpx/vpx_dsp/sum_squares.c', + 'libvpx/vpx_dsp/variance.c', + 'libvpx/vpx_dsp/vpx_convolve.c', + 'libvpx/vpx_dsp/vpx_dsp_rtcd.c', + 'libvpx/vpx_dsp/x86/add_noise_sse2.asm', + 'libvpx/vpx_dsp/x86/avg_intrin_avx2.c', + 'libvpx/vpx_dsp/x86/avg_intrin_sse2.c', + 'libvpx/vpx_dsp/x86/avg_pred_avx2.c', + 'libvpx/vpx_dsp/x86/avg_pred_sse2.c', + 'libvpx/vpx_dsp/x86/deblock_sse2.asm', + 'libvpx/vpx_dsp/x86/fwd_txfm_avx2.c', + 'libvpx/vpx_dsp/x86/fwd_txfm_sse2.c', + 'libvpx/vpx_dsp/x86/intrapred_sse2.asm', + 'libvpx/vpx_dsp/x86/intrapred_ssse3.asm', + 'libvpx/vpx_dsp/x86/inv_txfm_avx2.c', + 'libvpx/vpx_dsp/x86/inv_txfm_sse2.c', + 'libvpx/vpx_dsp/x86/inv_txfm_ssse3.c', + 'libvpx/vpx_dsp/x86/inv_wht_sse2.asm', + 'libvpx/vpx_dsp/x86/loopfilter_avx2.c', + 'libvpx/vpx_dsp/x86/loopfilter_intrin_sse2.c', + 'libvpx/vpx_dsp/x86/post_proc_sse2.c', + 'libvpx/vpx_dsp/x86/quantize_avx.c', + 'libvpx/vpx_dsp/x86/quantize_avx2.c', + 'libvpx/vpx_dsp/x86/quantize_sse2.c', + 'libvpx/vpx_dsp/x86/quantize_ssse3.c', + 'libvpx/vpx_dsp/x86/sad4d_avx2.c', + 'libvpx/vpx_dsp/x86/sad4d_sse2.asm', + 'libvpx/vpx_dsp/x86/sad_avx2.c', + 'libvpx/vpx_dsp/x86/sad_sse2.asm', + 'libvpx/vpx_dsp/x86/sse_avx2.c', + 'libvpx/vpx_dsp/x86/sse_sse4.c', + 'libvpx/vpx_dsp/x86/subpel_variance_sse2.asm', + 'libvpx/vpx_dsp/x86/subtract_avx2.c', + 'libvpx/vpx_dsp/x86/subtract_sse2.asm', + 'libvpx/vpx_dsp/x86/sum_squares_sse2.c', + 'libvpx/vpx_dsp/x86/variance_avx2.c', + 'libvpx/vpx_dsp/x86/variance_sse2.c', + 'libvpx/vpx_dsp/x86/vpx_convolve_copy_sse2.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_4t_intrin_sse2.c', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_intrin_avx2.c', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_intrin_ssse3.c', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_sse2.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_ssse3.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_bilinear_sse2.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_bilinear_ssse3.asm', + 'libvpx/vpx_mem/vpx_mem.c', + 'libvpx/vpx_ports/emms_mmx.c', + 'libvpx/vpx_scale/generic/gen_scalers.c', + 'libvpx/vpx_scale/generic/vpx_scale.c', + 'libvpx/vpx_scale/generic/yv12config.c', + 'libvpx/vpx_scale/generic/yv12extend.c', + 'libvpx/vpx_scale/vpx_scale_rtcd.c', + 'libvpx/vpx_util/vpx_thread.c', + 'libvpx/vpx_util/vpx_write_yuv_frame.c', +], + 'WIN_IA32_EXPORTS': [ + 'libvpx/vpx/vp8.h', + 'libvpx/vpx/vp8cx.h', + 'libvpx/vpx/vp8dx.h', + 'libvpx/vpx/vpx_codec.h', + 'libvpx/vpx/vpx_decoder.h', + 'libvpx/vpx/vpx_encoder.h', + 'libvpx/vpx/vpx_ext_ratectrl.h', + 'libvpx/vpx/vpx_frame_buffer.h', + 'libvpx/vpx/vpx_image.h', + 'libvpx/vpx/vpx_integer.h', + 'libvpx/vpx/vpx_tpl.h', + 'libvpx/vpx_mem/include/vpx_mem_intrnl.h', + 'libvpx/vpx_mem/vpx_mem.h', + 'libvpx/vpx_ports/bitops.h', + 'libvpx/vpx_ports/compiler_attributes.h', + 'libvpx/vpx_ports/mem.h', + 'libvpx/vpx_ports/static_assert.h', + 'libvpx/vpx_ports/system_state.h', + 'libvpx/vpx_ports/vpx_timer.h', + 'libvpx/vpx_ports/x86.h', + 'libvpx/vpx_scale/vpx_scale.h', + 'libvpx/vpx_scale/yv12config.h', +], + 'WIN_IA32_SOURCES': [ + 'libvpx/../config/win/ia32/vpx_config.c', + 'libvpx/vp8/common/alloccommon.c', + 'libvpx/vp8/common/blockd.c', + 'libvpx/vp8/common/dequantize.c', + 'libvpx/vp8/common/entropy.c', + 'libvpx/vp8/common/entropymode.c', + 'libvpx/vp8/common/entropymv.c', + 'libvpx/vp8/common/extend.c', + 'libvpx/vp8/common/filter.c', + 'libvpx/vp8/common/findnearmv.c', + 'libvpx/vp8/common/generic/systemdependent.c', + 'libvpx/vp8/common/idct_blk.c', + 'libvpx/vp8/common/idctllm.c', + 'libvpx/vp8/common/loopfilter_filters.c', + 'libvpx/vp8/common/mbpitch.c', + 'libvpx/vp8/common/mfqe.c', + 'libvpx/vp8/common/modecont.c', + 'libvpx/vp8/common/postproc.c', + 'libvpx/vp8/common/quant_common.c', + 'libvpx/vp8/common/reconinter.c', + 'libvpx/vp8/common/reconintra.c', + 'libvpx/vp8/common/reconintra4x4.c', + 'libvpx/vp8/common/rtcd.c', + 'libvpx/vp8/common/setupintrarecon.c', + 'libvpx/vp8/common/swapyv12buffer.c', + 'libvpx/vp8/common/treecoder.c', + 'libvpx/vp8/common/vp8_loopfilter.c', + 'libvpx/vp8/common/vp8_skin_detection.c', + 'libvpx/vp8/common/x86/bilinear_filter_sse2.c', + 'libvpx/vp8/common/x86/dequantize_mmx.asm', + 'libvpx/vp8/common/x86/idct_blk_mmx.c', + 'libvpx/vp8/common/x86/idct_blk_sse2.c', + 'libvpx/vp8/common/x86/idctllm_mmx.asm', + 'libvpx/vp8/common/x86/idctllm_sse2.asm', + 'libvpx/vp8/common/x86/iwalsh_sse2.asm', + 'libvpx/vp8/common/x86/loopfilter_sse2.asm', + 'libvpx/vp8/common/x86/loopfilter_x86.c', + 'libvpx/vp8/common/x86/mfqe_sse2.asm', + 'libvpx/vp8/common/x86/recon_mmx.asm', + 'libvpx/vp8/common/x86/recon_sse2.asm', + 'libvpx/vp8/common/x86/subpixel_mmx.asm', + 'libvpx/vp8/common/x86/subpixel_sse2.asm', + 'libvpx/vp8/common/x86/subpixel_ssse3.asm', + 'libvpx/vp8/common/x86/vp8_asm_stubs.c', + 'libvpx/vp8/decoder/dboolhuff.c', + 'libvpx/vp8/decoder/decodeframe.c', + 'libvpx/vp8/decoder/decodemv.c', + 'libvpx/vp8/decoder/detokenize.c', + 'libvpx/vp8/decoder/onyxd_if.c', + 'libvpx/vp8/decoder/threading.c', + 'libvpx/vp8/encoder/bitstream.c', + 'libvpx/vp8/encoder/boolhuff.c', + 'libvpx/vp8/encoder/copy_c.c', + 'libvpx/vp8/encoder/dct.c', + 'libvpx/vp8/encoder/denoising.c', + 'libvpx/vp8/encoder/encodeframe.c', + 'libvpx/vp8/encoder/encodeintra.c', + 'libvpx/vp8/encoder/encodemb.c', + 'libvpx/vp8/encoder/encodemv.c', + 'libvpx/vp8/encoder/ethreading.c', + 'libvpx/vp8/encoder/firstpass.c', + 'libvpx/vp8/encoder/lookahead.c', + 'libvpx/vp8/encoder/mcomp.c', + 'libvpx/vp8/encoder/modecosts.c', + 'libvpx/vp8/encoder/mr_dissim.c', + 'libvpx/vp8/encoder/onyx_if.c', + 'libvpx/vp8/encoder/pickinter.c', + 'libvpx/vp8/encoder/picklpf.c', + 'libvpx/vp8/encoder/ratectrl.c', + 'libvpx/vp8/encoder/rdopt.c', + 'libvpx/vp8/encoder/segmentation.c', + 'libvpx/vp8/encoder/temporal_filter.c', + 'libvpx/vp8/encoder/tokenize.c', + 'libvpx/vp8/encoder/treewriter.c', + 'libvpx/vp8/encoder/vp8_quantize.c', + 'libvpx/vp8/encoder/x86/block_error_sse2.asm', + 'libvpx/vp8/encoder/x86/copy_sse2.asm', + 'libvpx/vp8/encoder/x86/copy_sse3.asm', + 'libvpx/vp8/encoder/x86/dct_sse2.asm', + 'libvpx/vp8/encoder/x86/denoising_sse2.c', + 'libvpx/vp8/encoder/x86/fwalsh_sse2.asm', + 'libvpx/vp8/encoder/x86/quantize_sse4.c', + 'libvpx/vp8/encoder/x86/temporal_filter_apply_sse2.asm', + 'libvpx/vp8/encoder/x86/vp8_enc_stubs_sse2.c', + 'libvpx/vp8/encoder/x86/vp8_quantize_sse2.c', + 'libvpx/vp8/encoder/x86/vp8_quantize_ssse3.c', + 'libvpx/vp8/vp8_cx_iface.c', + 'libvpx/vp8/vp8_dx_iface.c', + 'libvpx/vp9/common/vp9_alloccommon.c', + 'libvpx/vp9/common/vp9_blockd.c', + 'libvpx/vp9/common/vp9_common_data.c', + 'libvpx/vp9/common/vp9_entropy.c', + 'libvpx/vp9/common/vp9_entropymode.c', + 'libvpx/vp9/common/vp9_entropymv.c', + 'libvpx/vp9/common/vp9_filter.c', + 'libvpx/vp9/common/vp9_frame_buffers.c', + 'libvpx/vp9/common/vp9_idct.c', + 'libvpx/vp9/common/vp9_loopfilter.c', + 'libvpx/vp9/common/vp9_mfqe.c', + 'libvpx/vp9/common/vp9_mvref_common.c', + 'libvpx/vp9/common/vp9_postproc.c', + 'libvpx/vp9/common/vp9_pred_common.c', + 'libvpx/vp9/common/vp9_quant_common.c', + 'libvpx/vp9/common/vp9_reconinter.c', + 'libvpx/vp9/common/vp9_reconintra.c', + 'libvpx/vp9/common/vp9_rtcd.c', + 'libvpx/vp9/common/vp9_scale.c', + 'libvpx/vp9/common/vp9_scan.c', + 'libvpx/vp9/common/vp9_seg_common.c', + 'libvpx/vp9/common/vp9_thread_common.c', + 'libvpx/vp9/common/vp9_tile_common.c', + 'libvpx/vp9/common/x86/vp9_idct_intrin_sse2.c', + 'libvpx/vp9/common/x86/vp9_mfqe_sse2.asm', + 'libvpx/vp9/decoder/vp9_decodeframe.c', + 'libvpx/vp9/decoder/vp9_decodemv.c', + 'libvpx/vp9/decoder/vp9_decoder.c', + 'libvpx/vp9/decoder/vp9_detokenize.c', + 'libvpx/vp9/decoder/vp9_dsubexp.c', + 'libvpx/vp9/decoder/vp9_job_queue.c', + 'libvpx/vp9/encoder/vp9_alt_ref_aq.c', + 'libvpx/vp9/encoder/vp9_aq_360.c', + 'libvpx/vp9/encoder/vp9_aq_complexity.c', + 'libvpx/vp9/encoder/vp9_aq_cyclicrefresh.c', + 'libvpx/vp9/encoder/vp9_aq_variance.c', + 'libvpx/vp9/encoder/vp9_bitstream.c', + 'libvpx/vp9/encoder/vp9_context_tree.c', + 'libvpx/vp9/encoder/vp9_cost.c', + 'libvpx/vp9/encoder/vp9_dct.c', + 'libvpx/vp9/encoder/vp9_encodeframe.c', + 'libvpx/vp9/encoder/vp9_encodemb.c', + 'libvpx/vp9/encoder/vp9_encodemv.c', + 'libvpx/vp9/encoder/vp9_encoder.c', + 'libvpx/vp9/encoder/vp9_ethread.c', + 'libvpx/vp9/encoder/vp9_ext_ratectrl.c', + 'libvpx/vp9/encoder/vp9_extend.c', + 'libvpx/vp9/encoder/vp9_firstpass.c', + 'libvpx/vp9/encoder/vp9_frame_scale.c', + 'libvpx/vp9/encoder/vp9_lookahead.c', + 'libvpx/vp9/encoder/vp9_mbgraph.c', + 'libvpx/vp9/encoder/vp9_mcomp.c', + 'libvpx/vp9/encoder/vp9_multi_thread.c', + 'libvpx/vp9/encoder/vp9_noise_estimate.c', + 'libvpx/vp9/encoder/vp9_picklpf.c', + 'libvpx/vp9/encoder/vp9_pickmode.c', + 'libvpx/vp9/encoder/vp9_quantize.c', + 'libvpx/vp9/encoder/vp9_ratectrl.c', + 'libvpx/vp9/encoder/vp9_rd.c', + 'libvpx/vp9/encoder/vp9_rdopt.c', + 'libvpx/vp9/encoder/vp9_resize.c', + 'libvpx/vp9/encoder/vp9_segmentation.c', + 'libvpx/vp9/encoder/vp9_skin_detection.c', + 'libvpx/vp9/encoder/vp9_speed_features.c', + 'libvpx/vp9/encoder/vp9_subexp.c', + 'libvpx/vp9/encoder/vp9_svc_layercontext.c', + 'libvpx/vp9/encoder/vp9_temporal_filter.c', + 'libvpx/vp9/encoder/vp9_tokenize.c', + 'libvpx/vp9/encoder/vp9_tpl_model.c', + 'libvpx/vp9/encoder/vp9_treewriter.c', + 'libvpx/vp9/encoder/x86/temporal_filter_sse4.c', + 'libvpx/vp9/encoder/x86/vp9_dct_intrin_sse2.c', + 'libvpx/vp9/encoder/x86/vp9_dct_sse2.asm', + 'libvpx/vp9/encoder/x86/vp9_error_avx2.c', + 'libvpx/vp9/encoder/x86/vp9_error_sse2.asm', + 'libvpx/vp9/encoder/x86/vp9_frame_scale_ssse3.c', + 'libvpx/vp9/encoder/x86/vp9_quantize_avx2.c', + 'libvpx/vp9/encoder/x86/vp9_quantize_sse2.c', + 'libvpx/vp9/encoder/x86/vp9_quantize_ssse3.c', + 'libvpx/vp9/vp9_cx_iface.c', + 'libvpx/vp9/vp9_dx_iface.c', + 'libvpx/vp9/vp9_iface_common.c', + 'libvpx/vpx/src/vpx_codec.c', + 'libvpx/vpx/src/vpx_decoder.c', + 'libvpx/vpx/src/vpx_encoder.c', + 'libvpx/vpx/src/vpx_image.c', + 'libvpx/vpx_dsp/add_noise.c', + 'libvpx/vpx_dsp/avg.c', + 'libvpx/vpx_dsp/bitreader.c', + 'libvpx/vpx_dsp/bitreader_buffer.c', + 'libvpx/vpx_dsp/bitwriter.c', + 'libvpx/vpx_dsp/bitwriter_buffer.c', + 'libvpx/vpx_dsp/deblock.c', + 'libvpx/vpx_dsp/fwd_txfm.c', + 'libvpx/vpx_dsp/intrapred.c', + 'libvpx/vpx_dsp/inv_txfm.c', + 'libvpx/vpx_dsp/loopfilter.c', + 'libvpx/vpx_dsp/prob.c', + 'libvpx/vpx_dsp/psnr.c', + 'libvpx/vpx_dsp/quantize.c', + 'libvpx/vpx_dsp/sad.c', + 'libvpx/vpx_dsp/skin_detection.c', + 'libvpx/vpx_dsp/sse.c', + 'libvpx/vpx_dsp/subtract.c', + 'libvpx/vpx_dsp/sum_squares.c', + 'libvpx/vpx_dsp/variance.c', + 'libvpx/vpx_dsp/vpx_convolve.c', + 'libvpx/vpx_dsp/vpx_dsp_rtcd.c', + 'libvpx/vpx_dsp/x86/add_noise_sse2.asm', + 'libvpx/vpx_dsp/x86/avg_intrin_avx2.c', + 'libvpx/vpx_dsp/x86/avg_intrin_sse2.c', + 'libvpx/vpx_dsp/x86/avg_pred_avx2.c', + 'libvpx/vpx_dsp/x86/avg_pred_sse2.c', + 'libvpx/vpx_dsp/x86/deblock_sse2.asm', + 'libvpx/vpx_dsp/x86/fwd_txfm_avx2.c', + 'libvpx/vpx_dsp/x86/fwd_txfm_sse2.c', + 'libvpx/vpx_dsp/x86/intrapred_sse2.asm', + 'libvpx/vpx_dsp/x86/intrapred_ssse3.asm', + 'libvpx/vpx_dsp/x86/inv_txfm_avx2.c', + 'libvpx/vpx_dsp/x86/inv_txfm_sse2.c', + 'libvpx/vpx_dsp/x86/inv_txfm_ssse3.c', + 'libvpx/vpx_dsp/x86/inv_wht_sse2.asm', + 'libvpx/vpx_dsp/x86/loopfilter_avx2.c', + 'libvpx/vpx_dsp/x86/loopfilter_intrin_sse2.c', + 'libvpx/vpx_dsp/x86/post_proc_sse2.c', + 'libvpx/vpx_dsp/x86/quantize_avx.c', + 'libvpx/vpx_dsp/x86/quantize_avx2.c', + 'libvpx/vpx_dsp/x86/quantize_sse2.c', + 'libvpx/vpx_dsp/x86/quantize_ssse3.c', + 'libvpx/vpx_dsp/x86/sad4d_avx2.c', + 'libvpx/vpx_dsp/x86/sad4d_sse2.asm', + 'libvpx/vpx_dsp/x86/sad_avx2.c', + 'libvpx/vpx_dsp/x86/sad_sse2.asm', + 'libvpx/vpx_dsp/x86/sse_avx2.c', + 'libvpx/vpx_dsp/x86/sse_sse4.c', + 'libvpx/vpx_dsp/x86/subpel_variance_sse2.asm', + 'libvpx/vpx_dsp/x86/subtract_avx2.c', + 'libvpx/vpx_dsp/x86/subtract_sse2.asm', + 'libvpx/vpx_dsp/x86/sum_squares_sse2.c', + 'libvpx/vpx_dsp/x86/variance_avx2.c', + 'libvpx/vpx_dsp/x86/variance_sse2.c', + 'libvpx/vpx_dsp/x86/vpx_convolve_copy_sse2.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_4t_intrin_sse2.c', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_intrin_avx2.c', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_intrin_ssse3.c', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_sse2.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_8t_ssse3.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_bilinear_sse2.asm', + 'libvpx/vpx_dsp/x86/vpx_subpixel_bilinear_ssse3.asm', + 'libvpx/vpx_mem/vpx_mem.c', + 'libvpx/vpx_ports/emms_mmx.c', + 'libvpx/vpx_scale/generic/gen_scalers.c', + 'libvpx/vpx_scale/generic/vpx_scale.c', + 'libvpx/vpx_scale/generic/yv12config.c', + 'libvpx/vpx_scale/generic/yv12extend.c', + 'libvpx/vpx_scale/vpx_scale_rtcd.c', + 'libvpx/vpx_util/vpx_thread.c', + 'libvpx/vpx_util/vpx_write_yuv_frame.c', +], + 'LINUX_ARM_EXPORTS': [ 'libvpx/vpx/vp8.h', 'libvpx/vpx/vp8cx.h', 'libvpx/vpx/vp8dx.h', @@ -575,7 +1673,8 @@ files = { 'libvpx/vpx_scale/vpx_scale.h', 'libvpx/vpx_scale/yv12config.h', ], - 'ARM_SOURCES': [ + 'LINUX_ARM_SOURCES': [ + 'libvpx/../config/linux/arm/vpx_config.c', 'libvpx/vp8/common/alloccommon.c', 'libvpx/vp8/common/arm/loopfilter_arm.c', 'libvpx/vp8/common/arm/neon/bilinearpredict_neon.c', @@ -724,7 +1823,6 @@ files = { 'libvpx/vpx/src/vpx_decoder.c', 'libvpx/vpx/src/vpx_encoder.c', 'libvpx/vpx/src/vpx_image.c', - 'libvpx/vpx/src/vpx_tpl.c', 'libvpx/vpx_dsp/arm/avg_neon.c', 'libvpx/vpx_dsp/arm/avg_pred_neon.c', 'libvpx/vpx_dsp/arm/fdct16x16_neon.c', @@ -801,7 +1899,251 @@ files = { 'libvpx/vpx_util/vpx_thread.c', 'libvpx/vpx_util/vpx_write_yuv_frame.c', ], - 'ARM64_EXPORTS': [ + 'LINUX_ARM64_EXPORTS': [ + 'libvpx/vpx/vp8.h', + 'libvpx/vpx/vp8cx.h', + 'libvpx/vpx/vp8dx.h', + 'libvpx/vpx/vpx_codec.h', + 'libvpx/vpx/vpx_decoder.h', + 'libvpx/vpx/vpx_encoder.h', + 'libvpx/vpx/vpx_ext_ratectrl.h', + 'libvpx/vpx/vpx_frame_buffer.h', + 'libvpx/vpx/vpx_image.h', + 'libvpx/vpx/vpx_integer.h', + 'libvpx/vpx/vpx_tpl.h', + 'libvpx/vpx_mem/include/vpx_mem_intrnl.h', + 'libvpx/vpx_mem/vpx_mem.h', + 'libvpx/vpx_ports/arm.h', + 'libvpx/vpx_ports/arm_cpudetect.h', + 'libvpx/vpx_ports/bitops.h', + 'libvpx/vpx_ports/compiler_attributes.h', + 'libvpx/vpx_ports/mem.h', + 'libvpx/vpx_ports/static_assert.h', + 'libvpx/vpx_ports/system_state.h', + 'libvpx/vpx_ports/vpx_timer.h', + 'libvpx/vpx_scale/vpx_scale.h', + 'libvpx/vpx_scale/yv12config.h', +], + 'LINUX_ARM64_SOURCES': [ + 'libvpx/../config/linux/arm64/vpx_config.c', + 'libvpx/vp8/common/alloccommon.c', + 'libvpx/vp8/common/arm/loopfilter_arm.c', + 'libvpx/vp8/common/arm/neon/bilinearpredict_neon.c', + 'libvpx/vp8/common/arm/neon/copymem_neon.c', + 'libvpx/vp8/common/arm/neon/dc_only_idct_add_neon.c', + 'libvpx/vp8/common/arm/neon/dequant_idct_neon.c', + 'libvpx/vp8/common/arm/neon/dequantizeb_neon.c', + 'libvpx/vp8/common/arm/neon/idct_blk_neon.c', + 'libvpx/vp8/common/arm/neon/iwalsh_neon.c', + 'libvpx/vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.c', + 'libvpx/vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.c', + 'libvpx/vp8/common/arm/neon/mbloopfilter_neon.c', + 'libvpx/vp8/common/arm/neon/shortidct4x4llm_neon.c', + 'libvpx/vp8/common/arm/neon/sixtappredict_neon.c', + 'libvpx/vp8/common/arm/neon/vp8_loopfilter_neon.c', + 'libvpx/vp8/common/blockd.c', + 'libvpx/vp8/common/dequantize.c', + 'libvpx/vp8/common/entropy.c', + 'libvpx/vp8/common/entropymode.c', + 'libvpx/vp8/common/entropymv.c', + 'libvpx/vp8/common/extend.c', + 'libvpx/vp8/common/filter.c', + 'libvpx/vp8/common/findnearmv.c', + 'libvpx/vp8/common/generic/systemdependent.c', + 'libvpx/vp8/common/idct_blk.c', + 'libvpx/vp8/common/idctllm.c', + 'libvpx/vp8/common/loopfilter_filters.c', + 'libvpx/vp8/common/mbpitch.c', + 'libvpx/vp8/common/modecont.c', + 'libvpx/vp8/common/quant_common.c', + 'libvpx/vp8/common/reconinter.c', + 'libvpx/vp8/common/reconintra.c', + 'libvpx/vp8/common/reconintra4x4.c', + 'libvpx/vp8/common/rtcd.c', + 'libvpx/vp8/common/setupintrarecon.c', + 'libvpx/vp8/common/swapyv12buffer.c', + 'libvpx/vp8/common/treecoder.c', + 'libvpx/vp8/common/vp8_loopfilter.c', + 'libvpx/vp8/common/vp8_skin_detection.c', + 'libvpx/vp8/decoder/dboolhuff.c', + 'libvpx/vp8/decoder/decodeframe.c', + 'libvpx/vp8/decoder/decodemv.c', + 'libvpx/vp8/decoder/detokenize.c', + 'libvpx/vp8/decoder/onyxd_if.c', + 'libvpx/vp8/decoder/threading.c', + 'libvpx/vp8/encoder/arm/neon/denoising_neon.c', + 'libvpx/vp8/encoder/arm/neon/fastquantizeb_neon.c', + 'libvpx/vp8/encoder/arm/neon/shortfdct_neon.c', + 'libvpx/vp8/encoder/arm/neon/vp8_shortwalsh4x4_neon.c', + 'libvpx/vp8/encoder/bitstream.c', + 'libvpx/vp8/encoder/boolhuff.c', + 'libvpx/vp8/encoder/copy_c.c', + 'libvpx/vp8/encoder/dct.c', + 'libvpx/vp8/encoder/denoising.c', + 'libvpx/vp8/encoder/encodeframe.c', + 'libvpx/vp8/encoder/encodeintra.c', + 'libvpx/vp8/encoder/encodemb.c', + 'libvpx/vp8/encoder/encodemv.c', + 'libvpx/vp8/encoder/ethreading.c', + 'libvpx/vp8/encoder/lookahead.c', + 'libvpx/vp8/encoder/mcomp.c', + 'libvpx/vp8/encoder/modecosts.c', + 'libvpx/vp8/encoder/mr_dissim.c', + 'libvpx/vp8/encoder/onyx_if.c', + 'libvpx/vp8/encoder/pickinter.c', + 'libvpx/vp8/encoder/picklpf.c', + 'libvpx/vp8/encoder/ratectrl.c', + 'libvpx/vp8/encoder/rdopt.c', + 'libvpx/vp8/encoder/segmentation.c', + 'libvpx/vp8/encoder/tokenize.c', + 'libvpx/vp8/encoder/treewriter.c', + 'libvpx/vp8/encoder/vp8_quantize.c', + 'libvpx/vp8/vp8_cx_iface.c', + 'libvpx/vp8/vp8_dx_iface.c', + 'libvpx/vp9/common/arm/neon/vp9_iht16x16_add_neon.c', + 'libvpx/vp9/common/arm/neon/vp9_iht4x4_add_neon.c', + 'libvpx/vp9/common/arm/neon/vp9_iht8x8_add_neon.c', + 'libvpx/vp9/common/vp9_alloccommon.c', + 'libvpx/vp9/common/vp9_blockd.c', + 'libvpx/vp9/common/vp9_common_data.c', + 'libvpx/vp9/common/vp9_entropy.c', + 'libvpx/vp9/common/vp9_entropymode.c', + 'libvpx/vp9/common/vp9_entropymv.c', + 'libvpx/vp9/common/vp9_filter.c', + 'libvpx/vp9/common/vp9_frame_buffers.c', + 'libvpx/vp9/common/vp9_idct.c', + 'libvpx/vp9/common/vp9_loopfilter.c', + 'libvpx/vp9/common/vp9_mvref_common.c', + 'libvpx/vp9/common/vp9_pred_common.c', + 'libvpx/vp9/common/vp9_quant_common.c', + 'libvpx/vp9/common/vp9_reconinter.c', + 'libvpx/vp9/common/vp9_reconintra.c', + 'libvpx/vp9/common/vp9_rtcd.c', + 'libvpx/vp9/common/vp9_scale.c', + 'libvpx/vp9/common/vp9_scan.c', + 'libvpx/vp9/common/vp9_seg_common.c', + 'libvpx/vp9/common/vp9_thread_common.c', + 'libvpx/vp9/common/vp9_tile_common.c', + 'libvpx/vp9/decoder/vp9_decodeframe.c', + 'libvpx/vp9/decoder/vp9_decodemv.c', + 'libvpx/vp9/decoder/vp9_decoder.c', + 'libvpx/vp9/decoder/vp9_detokenize.c', + 'libvpx/vp9/decoder/vp9_dsubexp.c', + 'libvpx/vp9/decoder/vp9_job_queue.c', + 'libvpx/vp9/encoder/arm/neon/vp9_dct_neon.c', + 'libvpx/vp9/encoder/arm/neon/vp9_diamond_search_sad_neon.c', + 'libvpx/vp9/encoder/arm/neon/vp9_error_neon.c', + 'libvpx/vp9/encoder/arm/neon/vp9_frame_scale_neon.c', + 'libvpx/vp9/encoder/arm/neon/vp9_quantize_neon.c', + 'libvpx/vp9/encoder/vp9_aq_cyclicrefresh.c', + 'libvpx/vp9/encoder/vp9_bitstream.c', + 'libvpx/vp9/encoder/vp9_context_tree.c', + 'libvpx/vp9/encoder/vp9_cost.c', + 'libvpx/vp9/encoder/vp9_dct.c', + 'libvpx/vp9/encoder/vp9_encodeframe.c', + 'libvpx/vp9/encoder/vp9_encodemb.c', + 'libvpx/vp9/encoder/vp9_encodemv.c', + 'libvpx/vp9/encoder/vp9_encoder.c', + 'libvpx/vp9/encoder/vp9_ethread.c', + 'libvpx/vp9/encoder/vp9_ext_ratectrl.c', + 'libvpx/vp9/encoder/vp9_extend.c', + 'libvpx/vp9/encoder/vp9_frame_scale.c', + 'libvpx/vp9/encoder/vp9_lookahead.c', + 'libvpx/vp9/encoder/vp9_mcomp.c', + 'libvpx/vp9/encoder/vp9_multi_thread.c', + 'libvpx/vp9/encoder/vp9_noise_estimate.c', + 'libvpx/vp9/encoder/vp9_picklpf.c', + 'libvpx/vp9/encoder/vp9_pickmode.c', + 'libvpx/vp9/encoder/vp9_quantize.c', + 'libvpx/vp9/encoder/vp9_ratectrl.c', + 'libvpx/vp9/encoder/vp9_rd.c', + 'libvpx/vp9/encoder/vp9_rdopt.c', + 'libvpx/vp9/encoder/vp9_resize.c', + 'libvpx/vp9/encoder/vp9_segmentation.c', + 'libvpx/vp9/encoder/vp9_skin_detection.c', + 'libvpx/vp9/encoder/vp9_speed_features.c', + 'libvpx/vp9/encoder/vp9_subexp.c', + 'libvpx/vp9/encoder/vp9_svc_layercontext.c', + 'libvpx/vp9/encoder/vp9_tokenize.c', + 'libvpx/vp9/encoder/vp9_tpl_model.c', + 'libvpx/vp9/encoder/vp9_treewriter.c', + 'libvpx/vp9/vp9_cx_iface.c', + 'libvpx/vp9/vp9_dx_iface.c', + 'libvpx/vp9/vp9_iface_common.c', + 'libvpx/vpx/src/vpx_codec.c', + 'libvpx/vpx/src/vpx_decoder.c', + 'libvpx/vpx/src/vpx_encoder.c', + 'libvpx/vpx/src/vpx_image.c', + 'libvpx/vpx_dsp/arm/avg_neon.c', + 'libvpx/vpx_dsp/arm/avg_pred_neon.c', + 'libvpx/vpx_dsp/arm/fdct16x16_neon.c', + 'libvpx/vpx_dsp/arm/fdct32x32_neon.c', + 'libvpx/vpx_dsp/arm/fdct4x4_neon.c', + 'libvpx/vpx_dsp/arm/fdct8x8_neon.c', + 'libvpx/vpx_dsp/arm/fdct_partial_neon.c', + 'libvpx/vpx_dsp/arm/hadamard_neon.c', + 'libvpx/vpx_dsp/arm/idct16x16_1_add_neon.c', + 'libvpx/vpx_dsp/arm/idct16x16_add_neon.c', + 'libvpx/vpx_dsp/arm/idct32x32_135_add_neon.c', + 'libvpx/vpx_dsp/arm/idct32x32_1_add_neon.c', + 'libvpx/vpx_dsp/arm/idct32x32_34_add_neon.c', + 'libvpx/vpx_dsp/arm/idct32x32_add_neon.c', + 'libvpx/vpx_dsp/arm/idct4x4_1_add_neon.c', + 'libvpx/vpx_dsp/arm/idct4x4_add_neon.c', + 'libvpx/vpx_dsp/arm/idct8x8_1_add_neon.c', + 'libvpx/vpx_dsp/arm/idct8x8_add_neon.c', + 'libvpx/vpx_dsp/arm/intrapred_neon.c', + 'libvpx/vpx_dsp/arm/loopfilter_neon.c', + 'libvpx/vpx_dsp/arm/quantize_neon.c', + 'libvpx/vpx_dsp/arm/sad4d_neon.c', + 'libvpx/vpx_dsp/arm/sad4d_neon_dotprod.c', + 'libvpx/vpx_dsp/arm/sad_neon.c', + 'libvpx/vpx_dsp/arm/sad_neon_dotprod.c', + 'libvpx/vpx_dsp/arm/sse_neon.c', + 'libvpx/vpx_dsp/arm/sse_neon_dotprod.c', + 'libvpx/vpx_dsp/arm/subpel_variance_neon.c', + 'libvpx/vpx_dsp/arm/subtract_neon.c', + 'libvpx/vpx_dsp/arm/sum_squares_neon.c', + 'libvpx/vpx_dsp/arm/variance_neon.c', + 'libvpx/vpx_dsp/arm/variance_neon_dotprod.c', + 'libvpx/vpx_dsp/arm/vpx_convolve8_neon.c', + 'libvpx/vpx_dsp/arm/vpx_convolve8_neon_dotprod.c', + 'libvpx/vpx_dsp/arm/vpx_convolve8_neon_i8mm.c', + 'libvpx/vpx_dsp/arm/vpx_convolve_avg_neon.c', + 'libvpx/vpx_dsp/arm/vpx_convolve_copy_neon.c', + 'libvpx/vpx_dsp/arm/vpx_convolve_neon.c', + 'libvpx/vpx_dsp/arm/vpx_scaled_convolve8_neon.c', + 'libvpx/vpx_dsp/avg.c', + 'libvpx/vpx_dsp/bitreader.c', + 'libvpx/vpx_dsp/bitreader_buffer.c', + 'libvpx/vpx_dsp/bitwriter.c', + 'libvpx/vpx_dsp/bitwriter_buffer.c', + 'libvpx/vpx_dsp/fwd_txfm.c', + 'libvpx/vpx_dsp/intrapred.c', + 'libvpx/vpx_dsp/inv_txfm.c', + 'libvpx/vpx_dsp/loopfilter.c', + 'libvpx/vpx_dsp/prob.c', + 'libvpx/vpx_dsp/psnr.c', + 'libvpx/vpx_dsp/quantize.c', + 'libvpx/vpx_dsp/sad.c', + 'libvpx/vpx_dsp/skin_detection.c', + 'libvpx/vpx_dsp/sse.c', + 'libvpx/vpx_dsp/subtract.c', + 'libvpx/vpx_dsp/sum_squares.c', + 'libvpx/vpx_dsp/variance.c', + 'libvpx/vpx_dsp/vpx_convolve.c', + 'libvpx/vpx_dsp/vpx_dsp_rtcd.c', + 'libvpx/vpx_mem/vpx_mem.c', + 'libvpx/vpx_ports/aarch64_cpudetect.c', + 'libvpx/vpx_scale/generic/gen_scalers.c', + 'libvpx/vpx_scale/generic/vpx_scale.c', + 'libvpx/vpx_scale/generic/yv12config.c', + 'libvpx/vpx_scale/generic/yv12extend.c', + 'libvpx/vpx_scale/vpx_scale_rtcd.c', + 'libvpx/vpx_util/vpx_thread.c', + 'libvpx/vpx_util/vpx_write_yuv_frame.c', +], + 'MAC_ARM64_EXPORTS': [ 'libvpx/vpx/vp8.h', 'libvpx/vpx/vp8cx.h', 'libvpx/vpx/vp8dx.h', @@ -826,7 +2168,8 @@ files = { 'libvpx/vpx_scale/vpx_scale.h', 'libvpx/vpx_scale/yv12config.h', ], - 'ARM64_SOURCES': [ + 'MAC_ARM64_SOURCES': [ + 'libvpx/../config/mac/arm64/vpx_config.c', 'libvpx/vp8/common/alloccommon.c', 'libvpx/vp8/common/arm/loopfilter_arm.c', 'libvpx/vp8/common/arm/neon/bilinearpredict_neon.c', @@ -886,6 +2229,7 @@ files = { 'libvpx/vp8/encoder/encodemb.c', 'libvpx/vp8/encoder/encodemv.c', 'libvpx/vp8/encoder/ethreading.c', + 'libvpx/vp8/encoder/firstpass.c', 'libvpx/vp8/encoder/lookahead.c', 'libvpx/vp8/encoder/mcomp.c', 'libvpx/vp8/encoder/modecosts.c', @@ -896,6 +2240,7 @@ files = { 'libvpx/vp8/encoder/ratectrl.c', 'libvpx/vp8/encoder/rdopt.c', 'libvpx/vp8/encoder/segmentation.c', + 'libvpx/vp8/encoder/temporal_filter.c', 'libvpx/vp8/encoder/tokenize.c', 'libvpx/vp8/encoder/treewriter.c', 'libvpx/vp8/encoder/vp8_quantize.c', @@ -937,7 +2282,12 @@ files = { 'libvpx/vp9/encoder/arm/neon/vp9_error_sve.c', 'libvpx/vp9/encoder/arm/neon/vp9_frame_scale_neon.c', 'libvpx/vp9/encoder/arm/neon/vp9_quantize_neon.c', + 'libvpx/vp9/encoder/arm/neon/vp9_temporal_filter_neon.c', + 'libvpx/vp9/encoder/vp9_alt_ref_aq.c', + 'libvpx/vp9/encoder/vp9_aq_360.c', + 'libvpx/vp9/encoder/vp9_aq_complexity.c', 'libvpx/vp9/encoder/vp9_aq_cyclicrefresh.c', + 'libvpx/vp9/encoder/vp9_aq_variance.c', 'libvpx/vp9/encoder/vp9_bitstream.c', 'libvpx/vp9/encoder/vp9_context_tree.c', 'libvpx/vp9/encoder/vp9_cost.c', @@ -949,8 +2299,10 @@ files = { 'libvpx/vp9/encoder/vp9_ethread.c', 'libvpx/vp9/encoder/vp9_ext_ratectrl.c', 'libvpx/vp9/encoder/vp9_extend.c', + 'libvpx/vp9/encoder/vp9_firstpass.c', 'libvpx/vp9/encoder/vp9_frame_scale.c', 'libvpx/vp9/encoder/vp9_lookahead.c', + 'libvpx/vp9/encoder/vp9_mbgraph.c', 'libvpx/vp9/encoder/vp9_mcomp.c', 'libvpx/vp9/encoder/vp9_multi_thread.c', 'libvpx/vp9/encoder/vp9_noise_estimate.c', @@ -966,6 +2318,7 @@ files = { 'libvpx/vp9/encoder/vp9_speed_features.c', 'libvpx/vp9/encoder/vp9_subexp.c', 'libvpx/vp9/encoder/vp9_svc_layercontext.c', + 'libvpx/vp9/encoder/vp9_temporal_filter.c', 'libvpx/vp9/encoder/vp9_tokenize.c', 'libvpx/vp9/encoder/vp9_tpl_model.c', 'libvpx/vp9/encoder/vp9_treewriter.c', @@ -976,7 +2329,6 @@ files = { 'libvpx/vpx/src/vpx_decoder.c', 'libvpx/vpx/src/vpx_encoder.c', 'libvpx/vpx/src/vpx_image.c', - 'libvpx/vpx/src/vpx_tpl.c', 'libvpx/vpx_dsp/arm/avg_neon.c', 'libvpx/vpx_dsp/arm/avg_pred_neon.c', 'libvpx/vpx_dsp/arm/fdct16x16_neon.c', @@ -1047,6 +2399,250 @@ files = { 'libvpx/vpx_util/vpx_thread.c', 'libvpx/vpx_util/vpx_write_yuv_frame.c', ], + 'WIN_AARCH64_EXPORTS': [ + 'libvpx/vpx/vp8.h', + 'libvpx/vpx/vp8cx.h', + 'libvpx/vpx/vp8dx.h', + 'libvpx/vpx/vpx_codec.h', + 'libvpx/vpx/vpx_decoder.h', + 'libvpx/vpx/vpx_encoder.h', + 'libvpx/vpx/vpx_ext_ratectrl.h', + 'libvpx/vpx/vpx_frame_buffer.h', + 'libvpx/vpx/vpx_image.h', + 'libvpx/vpx/vpx_integer.h', + 'libvpx/vpx/vpx_tpl.h', + 'libvpx/vpx_mem/include/vpx_mem_intrnl.h', + 'libvpx/vpx_mem/vpx_mem.h', + 'libvpx/vpx_ports/arm.h', + 'libvpx/vpx_ports/arm_cpudetect.h', + 'libvpx/vpx_ports/bitops.h', + 'libvpx/vpx_ports/compiler_attributes.h', + 'libvpx/vpx_ports/mem.h', + 'libvpx/vpx_ports/static_assert.h', + 'libvpx/vpx_ports/system_state.h', + 'libvpx/vpx_ports/vpx_timer.h', + 'libvpx/vpx_scale/vpx_scale.h', + 'libvpx/vpx_scale/yv12config.h', +], + 'WIN_AARCH64_SOURCES': [ + 'libvpx/../config/win/aarch64/vpx_config.c', + 'libvpx/vp8/common/alloccommon.c', + 'libvpx/vp8/common/arm/loopfilter_arm.c', + 'libvpx/vp8/common/arm/neon/bilinearpredict_neon.c', + 'libvpx/vp8/common/arm/neon/copymem_neon.c', + 'libvpx/vp8/common/arm/neon/dc_only_idct_add_neon.c', + 'libvpx/vp8/common/arm/neon/dequant_idct_neon.c', + 'libvpx/vp8/common/arm/neon/dequantizeb_neon.c', + 'libvpx/vp8/common/arm/neon/idct_blk_neon.c', + 'libvpx/vp8/common/arm/neon/iwalsh_neon.c', + 'libvpx/vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.c', + 'libvpx/vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.c', + 'libvpx/vp8/common/arm/neon/mbloopfilter_neon.c', + 'libvpx/vp8/common/arm/neon/shortidct4x4llm_neon.c', + 'libvpx/vp8/common/arm/neon/sixtappredict_neon.c', + 'libvpx/vp8/common/arm/neon/vp8_loopfilter_neon.c', + 'libvpx/vp8/common/blockd.c', + 'libvpx/vp8/common/dequantize.c', + 'libvpx/vp8/common/entropy.c', + 'libvpx/vp8/common/entropymode.c', + 'libvpx/vp8/common/entropymv.c', + 'libvpx/vp8/common/extend.c', + 'libvpx/vp8/common/filter.c', + 'libvpx/vp8/common/findnearmv.c', + 'libvpx/vp8/common/generic/systemdependent.c', + 'libvpx/vp8/common/idct_blk.c', + 'libvpx/vp8/common/idctllm.c', + 'libvpx/vp8/common/loopfilter_filters.c', + 'libvpx/vp8/common/mbpitch.c', + 'libvpx/vp8/common/modecont.c', + 'libvpx/vp8/common/quant_common.c', + 'libvpx/vp8/common/reconinter.c', + 'libvpx/vp8/common/reconintra.c', + 'libvpx/vp8/common/reconintra4x4.c', + 'libvpx/vp8/common/rtcd.c', + 'libvpx/vp8/common/setupintrarecon.c', + 'libvpx/vp8/common/swapyv12buffer.c', + 'libvpx/vp8/common/treecoder.c', + 'libvpx/vp8/common/vp8_loopfilter.c', + 'libvpx/vp8/common/vp8_skin_detection.c', + 'libvpx/vp8/decoder/dboolhuff.c', + 'libvpx/vp8/decoder/decodeframe.c', + 'libvpx/vp8/decoder/decodemv.c', + 'libvpx/vp8/decoder/detokenize.c', + 'libvpx/vp8/decoder/onyxd_if.c', + 'libvpx/vp8/decoder/threading.c', + 'libvpx/vp8/encoder/arm/neon/denoising_neon.c', + 'libvpx/vp8/encoder/arm/neon/fastquantizeb_neon.c', + 'libvpx/vp8/encoder/arm/neon/shortfdct_neon.c', + 'libvpx/vp8/encoder/arm/neon/vp8_shortwalsh4x4_neon.c', + 'libvpx/vp8/encoder/bitstream.c', + 'libvpx/vp8/encoder/boolhuff.c', + 'libvpx/vp8/encoder/copy_c.c', + 'libvpx/vp8/encoder/dct.c', + 'libvpx/vp8/encoder/denoising.c', + 'libvpx/vp8/encoder/encodeframe.c', + 'libvpx/vp8/encoder/encodeintra.c', + 'libvpx/vp8/encoder/encodemb.c', + 'libvpx/vp8/encoder/encodemv.c', + 'libvpx/vp8/encoder/ethreading.c', + 'libvpx/vp8/encoder/lookahead.c', + 'libvpx/vp8/encoder/mcomp.c', + 'libvpx/vp8/encoder/modecosts.c', + 'libvpx/vp8/encoder/mr_dissim.c', + 'libvpx/vp8/encoder/onyx_if.c', + 'libvpx/vp8/encoder/pickinter.c', + 'libvpx/vp8/encoder/picklpf.c', + 'libvpx/vp8/encoder/ratectrl.c', + 'libvpx/vp8/encoder/rdopt.c', + 'libvpx/vp8/encoder/segmentation.c', + 'libvpx/vp8/encoder/tokenize.c', + 'libvpx/vp8/encoder/treewriter.c', + 'libvpx/vp8/encoder/vp8_quantize.c', + 'libvpx/vp8/vp8_cx_iface.c', + 'libvpx/vp8/vp8_dx_iface.c', + 'libvpx/vp9/common/arm/neon/vp9_iht16x16_add_neon.c', + 'libvpx/vp9/common/arm/neon/vp9_iht4x4_add_neon.c', + 'libvpx/vp9/common/arm/neon/vp9_iht8x8_add_neon.c', + 'libvpx/vp9/common/vp9_alloccommon.c', + 'libvpx/vp9/common/vp9_blockd.c', + 'libvpx/vp9/common/vp9_common_data.c', + 'libvpx/vp9/common/vp9_entropy.c', + 'libvpx/vp9/common/vp9_entropymode.c', + 'libvpx/vp9/common/vp9_entropymv.c', + 'libvpx/vp9/common/vp9_filter.c', + 'libvpx/vp9/common/vp9_frame_buffers.c', + 'libvpx/vp9/common/vp9_idct.c', + 'libvpx/vp9/common/vp9_loopfilter.c', + 'libvpx/vp9/common/vp9_mvref_common.c', + 'libvpx/vp9/common/vp9_pred_common.c', + 'libvpx/vp9/common/vp9_quant_common.c', + 'libvpx/vp9/common/vp9_reconinter.c', + 'libvpx/vp9/common/vp9_reconintra.c', + 'libvpx/vp9/common/vp9_rtcd.c', + 'libvpx/vp9/common/vp9_scale.c', + 'libvpx/vp9/common/vp9_scan.c', + 'libvpx/vp9/common/vp9_seg_common.c', + 'libvpx/vp9/common/vp9_thread_common.c', + 'libvpx/vp9/common/vp9_tile_common.c', + 'libvpx/vp9/decoder/vp9_decodeframe.c', + 'libvpx/vp9/decoder/vp9_decodemv.c', + 'libvpx/vp9/decoder/vp9_decoder.c', + 'libvpx/vp9/decoder/vp9_detokenize.c', + 'libvpx/vp9/decoder/vp9_dsubexp.c', + 'libvpx/vp9/decoder/vp9_job_queue.c', + 'libvpx/vp9/encoder/arm/neon/vp9_dct_neon.c', + 'libvpx/vp9/encoder/arm/neon/vp9_diamond_search_sad_neon.c', + 'libvpx/vp9/encoder/arm/neon/vp9_error_neon.c', + 'libvpx/vp9/encoder/arm/neon/vp9_frame_scale_neon.c', + 'libvpx/vp9/encoder/arm/neon/vp9_quantize_neon.c', + 'libvpx/vp9/encoder/vp9_aq_cyclicrefresh.c', + 'libvpx/vp9/encoder/vp9_bitstream.c', + 'libvpx/vp9/encoder/vp9_context_tree.c', + 'libvpx/vp9/encoder/vp9_cost.c', + 'libvpx/vp9/encoder/vp9_dct.c', + 'libvpx/vp9/encoder/vp9_encodeframe.c', + 'libvpx/vp9/encoder/vp9_encodemb.c', + 'libvpx/vp9/encoder/vp9_encodemv.c', + 'libvpx/vp9/encoder/vp9_encoder.c', + 'libvpx/vp9/encoder/vp9_ethread.c', + 'libvpx/vp9/encoder/vp9_ext_ratectrl.c', + 'libvpx/vp9/encoder/vp9_extend.c', + 'libvpx/vp9/encoder/vp9_frame_scale.c', + 'libvpx/vp9/encoder/vp9_lookahead.c', + 'libvpx/vp9/encoder/vp9_mcomp.c', + 'libvpx/vp9/encoder/vp9_multi_thread.c', + 'libvpx/vp9/encoder/vp9_noise_estimate.c', + 'libvpx/vp9/encoder/vp9_picklpf.c', + 'libvpx/vp9/encoder/vp9_pickmode.c', + 'libvpx/vp9/encoder/vp9_quantize.c', + 'libvpx/vp9/encoder/vp9_ratectrl.c', + 'libvpx/vp9/encoder/vp9_rd.c', + 'libvpx/vp9/encoder/vp9_rdopt.c', + 'libvpx/vp9/encoder/vp9_resize.c', + 'libvpx/vp9/encoder/vp9_segmentation.c', + 'libvpx/vp9/encoder/vp9_skin_detection.c', + 'libvpx/vp9/encoder/vp9_speed_features.c', + 'libvpx/vp9/encoder/vp9_subexp.c', + 'libvpx/vp9/encoder/vp9_svc_layercontext.c', + 'libvpx/vp9/encoder/vp9_tokenize.c', + 'libvpx/vp9/encoder/vp9_tpl_model.c', + 'libvpx/vp9/encoder/vp9_treewriter.c', + 'libvpx/vp9/vp9_cx_iface.c', + 'libvpx/vp9/vp9_dx_iface.c', + 'libvpx/vp9/vp9_iface_common.c', + 'libvpx/vpx/src/vpx_codec.c', + 'libvpx/vpx/src/vpx_decoder.c', + 'libvpx/vpx/src/vpx_encoder.c', + 'libvpx/vpx/src/vpx_image.c', + 'libvpx/vpx_dsp/arm/avg_neon.c', + 'libvpx/vpx_dsp/arm/avg_pred_neon.c', + 'libvpx/vpx_dsp/arm/fdct16x16_neon.c', + 'libvpx/vpx_dsp/arm/fdct32x32_neon.c', + 'libvpx/vpx_dsp/arm/fdct4x4_neon.c', + 'libvpx/vpx_dsp/arm/fdct8x8_neon.c', + 'libvpx/vpx_dsp/arm/fdct_partial_neon.c', + 'libvpx/vpx_dsp/arm/hadamard_neon.c', + 'libvpx/vpx_dsp/arm/idct16x16_1_add_neon.c', + 'libvpx/vpx_dsp/arm/idct16x16_add_neon.c', + 'libvpx/vpx_dsp/arm/idct32x32_135_add_neon.c', + 'libvpx/vpx_dsp/arm/idct32x32_1_add_neon.c', + 'libvpx/vpx_dsp/arm/idct32x32_34_add_neon.c', + 'libvpx/vpx_dsp/arm/idct32x32_add_neon.c', + 'libvpx/vpx_dsp/arm/idct4x4_1_add_neon.c', + 'libvpx/vpx_dsp/arm/idct4x4_add_neon.c', + 'libvpx/vpx_dsp/arm/idct8x8_1_add_neon.c', + 'libvpx/vpx_dsp/arm/idct8x8_add_neon.c', + 'libvpx/vpx_dsp/arm/intrapred_neon.c', + 'libvpx/vpx_dsp/arm/loopfilter_neon.c', + 'libvpx/vpx_dsp/arm/quantize_neon.c', + 'libvpx/vpx_dsp/arm/sad4d_neon.c', + 'libvpx/vpx_dsp/arm/sad4d_neon_dotprod.c', + 'libvpx/vpx_dsp/arm/sad_neon.c', + 'libvpx/vpx_dsp/arm/sad_neon_dotprod.c', + 'libvpx/vpx_dsp/arm/sse_neon.c', + 'libvpx/vpx_dsp/arm/sse_neon_dotprod.c', + 'libvpx/vpx_dsp/arm/subpel_variance_neon.c', + 'libvpx/vpx_dsp/arm/subtract_neon.c', + 'libvpx/vpx_dsp/arm/sum_squares_neon.c', + 'libvpx/vpx_dsp/arm/variance_neon.c', + 'libvpx/vpx_dsp/arm/variance_neon_dotprod.c', + 'libvpx/vpx_dsp/arm/vpx_convolve8_neon.c', + 'libvpx/vpx_dsp/arm/vpx_convolve8_neon_dotprod.c', + 'libvpx/vpx_dsp/arm/vpx_convolve8_neon_i8mm.c', + 'libvpx/vpx_dsp/arm/vpx_convolve_avg_neon.c', + 'libvpx/vpx_dsp/arm/vpx_convolve_copy_neon.c', + 'libvpx/vpx_dsp/arm/vpx_convolve_neon.c', + 'libvpx/vpx_dsp/arm/vpx_scaled_convolve8_neon.c', + 'libvpx/vpx_dsp/avg.c', + 'libvpx/vpx_dsp/bitreader.c', + 'libvpx/vpx_dsp/bitreader_buffer.c', + 'libvpx/vpx_dsp/bitwriter.c', + 'libvpx/vpx_dsp/bitwriter_buffer.c', + 'libvpx/vpx_dsp/fwd_txfm.c', + 'libvpx/vpx_dsp/intrapred.c', + 'libvpx/vpx_dsp/inv_txfm.c', + 'libvpx/vpx_dsp/loopfilter.c', + 'libvpx/vpx_dsp/prob.c', + 'libvpx/vpx_dsp/psnr.c', + 'libvpx/vpx_dsp/quantize.c', + 'libvpx/vpx_dsp/sad.c', + 'libvpx/vpx_dsp/skin_detection.c', + 'libvpx/vpx_dsp/sse.c', + 'libvpx/vpx_dsp/subtract.c', + 'libvpx/vpx_dsp/sum_squares.c', + 'libvpx/vpx_dsp/variance.c', + 'libvpx/vpx_dsp/vpx_convolve.c', + 'libvpx/vpx_dsp/vpx_dsp_rtcd.c', + 'libvpx/vpx_mem/vpx_mem.c', + 'libvpx/vpx_ports/aarch64_cpudetect.c', + 'libvpx/vpx_scale/generic/gen_scalers.c', + 'libvpx/vpx_scale/generic/vpx_scale.c', + 'libvpx/vpx_scale/generic/yv12config.c', + 'libvpx/vpx_scale/generic/yv12extend.c', + 'libvpx/vpx_scale/vpx_scale_rtcd.c', + 'libvpx/vpx_util/vpx_thread.c', + 'libvpx/vpx_util/vpx_write_yuv_frame.c', +], 'GENERIC_EXPORTS': [ 'libvpx/vpx/vp8.h', 'libvpx/vpx/vp8cx.h', @@ -1071,6 +2667,7 @@ files = { 'libvpx/vpx_scale/yv12config.h', ], 'GENERIC_SOURCES': [ + 'libvpx/../config/generic/vpx_config.c', 'libvpx/vp8/common/alloccommon.c', 'libvpx/vp8/common/blockd.c', 'libvpx/vp8/common/dequantize.c', @@ -1202,7 +2799,6 @@ files = { 'libvpx/vpx/src/vpx_decoder.c', 'libvpx/vpx/src/vpx_encoder.c', 'libvpx/vpx/src/vpx_image.c', - 'libvpx/vpx/src/vpx_tpl.c', 'libvpx/vpx_dsp/avg.c', 'libvpx/vpx_dsp/bitreader.c', 'libvpx/vpx_dsp/bitreader_buffer.c', |