diff options
Diffstat (limited to '')
| -rw-r--r-- | third_party/aom/av1/common/arm/highbd_warp_plane_sve.c | 32 |
1 files changed, 17 insertions, 15 deletions
diff --git a/third_party/aom/av1/common/arm/highbd_warp_plane_sve.c b/third_party/aom/av1/common/arm/highbd_warp_plane_sve.c index 7a14f21846..87e033fd00 100644 --- a/third_party/aom/av1/common/arm/highbd_warp_plane_sve.c +++ b/third_party/aom/av1/common/arm/highbd_warp_plane_sve.c @@ -15,7 +15,7 @@ #include <arm_neon_sve_bridge.h> #include "aom_dsp/aom_dsp_common.h" -#include "aom_dsp/arm/dot_sve.h" +#include "aom_dsp/arm/aom_neon_sve_bridge.h" #include "aom_dsp/arm/mem_neon.h" #include "aom_dsp/arm/transpose_neon.h" #include "aom_ports/mem.h" @@ -24,8 +24,8 @@ #include "config/av1_rtcd.h" #include "highbd_warp_plane_neon.h" -static INLINE int16x8_t highbd_horizontal_filter_4x1_f4(uint16x8x2_t in, int bd, - int sx, int alpha) { +static AOM_FORCE_INLINE int16x8_t +highbd_horizontal_filter_4x1_f4(uint16x8x2_t in, int bd, int sx, int alpha) { int16x8_t f[4]; load_filters_4(f, sx, alpha); @@ -55,8 +55,8 @@ static INLINE int16x8_t highbd_horizontal_filter_4x1_f4(uint16x8x2_t in, int bd, return vcombine_s16(vmovn_s32(res), vdup_n_s16(0)); } -static INLINE int16x8_t highbd_horizontal_filter_8x1_f8(uint16x8x2_t in, int bd, - int sx, int alpha) { +static AOM_FORCE_INLINE int16x8_t +highbd_horizontal_filter_8x1_f8(uint16x8x2_t in, int bd, int sx, int alpha) { int16x8_t f[8]; load_filters_8(f, sx, alpha); @@ -103,8 +103,8 @@ static INLINE int16x8_t highbd_horizontal_filter_8x1_f8(uint16x8x2_t in, int bd, return vcombine_s16(vmovn_s32(res0), vmovn_s32(res1)); } -static INLINE int16x8_t highbd_horizontal_filter_4x1_f1(uint16x8x2_t in, int bd, - int sx) { +static AOM_FORCE_INLINE int16x8_t +highbd_horizontal_filter_4x1_f1(uint16x8x2_t in, int bd, int sx) { int16x8_t f = load_filters_1(sx); int16x8_t rv0 = vextq_s16(vreinterpretq_s16_u16(in.val[0]), @@ -133,8 +133,8 @@ static INLINE int16x8_t highbd_horizontal_filter_4x1_f1(uint16x8x2_t in, int bd, return vcombine_s16(vmovn_s32(res), vdup_n_s16(0)); } -static INLINE int16x8_t highbd_horizontal_filter_8x1_f1(uint16x8x2_t in, int bd, - int sx) { +static AOM_FORCE_INLINE int16x8_t +highbd_horizontal_filter_8x1_f1(uint16x8x2_t in, int bd, int sx) { int16x8_t f = load_filters_1(sx); int16x8_t rv0 = vextq_s16(vreinterpretq_s16_u16(in.val[0]), @@ -180,7 +180,8 @@ static INLINE int16x8_t highbd_horizontal_filter_8x1_f1(uint16x8x2_t in, int bd, return vcombine_s16(vmovn_s32(res0), vmovn_s32(res1)); } -static INLINE int32x4_t vertical_filter_4x1_f1(const int16x8_t *tmp, int sy) { +static AOM_FORCE_INLINE int32x4_t vertical_filter_4x1_f1(const int16x8_t *tmp, + int sy) { const int16x8_t f = load_filters_1(sy); const int16x4_t f0123 = vget_low_s16(f); const int16x4_t f4567 = vget_high_s16(f); @@ -197,7 +198,8 @@ static INLINE int32x4_t vertical_filter_4x1_f1(const int16x8_t *tmp, int sy) { return m0123; } -static INLINE int32x4x2_t vertical_filter_8x1_f1(const int16x8_t *tmp, int sy) { +static AOM_FORCE_INLINE int32x4x2_t vertical_filter_8x1_f1(const int16x8_t *tmp, + int sy) { const int16x8_t f = load_filters_1(sy); const int16x4_t f0123 = vget_low_s16(f); const int16x4_t f4567 = vget_high_s16(f); @@ -223,8 +225,8 @@ static INLINE int32x4x2_t vertical_filter_8x1_f1(const int16x8_t *tmp, int sy) { return (int32x4x2_t){ { m0123, m4567 } }; } -static INLINE int32x4_t vertical_filter_4x1_f4(const int16x8_t *tmp, int sy, - int gamma) { +static AOM_FORCE_INLINE int32x4_t vertical_filter_4x1_f4(const int16x8_t *tmp, + int sy, int gamma) { int16x8_t s0, s1, s2, s3; transpose_elems_s16_4x8( vget_low_s16(tmp[0]), vget_low_s16(tmp[1]), vget_low_s16(tmp[2]), @@ -244,8 +246,8 @@ static INLINE int32x4_t vertical_filter_4x1_f4(const int16x8_t *tmp, int sy, return vcombine_s32(vmovn_s64(m01), vmovn_s64(m23)); } -static INLINE int32x4x2_t vertical_filter_8x1_f8(const int16x8_t *tmp, int sy, - int gamma) { +static AOM_FORCE_INLINE int32x4x2_t vertical_filter_8x1_f8(const int16x8_t *tmp, + int sy, int gamma) { int16x8_t s0 = tmp[0]; int16x8_t s1 = tmp[1]; int16x8_t s2 = tmp[2]; |