diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-19 00:47:55 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-19 00:47:55 +0000 |
commit | 26a029d407be480d791972afb5975cf62c9360a6 (patch) | |
tree | f435a8308119effd964b339f76abb83a57c29483 /third_party/aom/av1/encoder/arm/neon/shift_neon.h | |
parent | Initial commit. (diff) | |
download | firefox-26a029d407be480d791972afb5975cf62c9360a6.tar.xz firefox-26a029d407be480d791972afb5975cf62c9360a6.zip |
Adding upstream version 124.0.1.upstream/124.0.1
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to '')
-rw-r--r-- | third_party/aom/av1/encoder/arm/neon/shift_neon.h | 49 |
1 files changed, 49 insertions, 0 deletions
diff --git a/third_party/aom/av1/encoder/arm/neon/shift_neon.h b/third_party/aom/av1/encoder/arm/neon/shift_neon.h new file mode 100644 index 0000000000..d73aef2f25 --- /dev/null +++ b/third_party/aom/av1/encoder/arm/neon/shift_neon.h @@ -0,0 +1,49 @@ +/* + * Copyright (c) 2023, Alliance for Open Media. All rights reserved + * + * This source code is subject to the terms of the BSD 2 Clause License and + * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License + * was not distributed with this source code in the LICENSE file, you can + * obtain it at www.aomedia.org/license/software. If the Alliance for Open + * Media Patent License 1.0 was not distributed with this source code in the + * PATENTS file, you can obtain it at www.aomedia.org/license/patent. + */ + +#ifndef AOM_AV1_ENCODER_ARM_NEON_SHIFT_NEON_H_ +#define AOM_AV1_ENCODER_ARM_NEON_SHIFT_NEON_H_ + +#include <arm_neon.h> + +#include "aom/aom_integer.h" // For AOM_INLINE. + +#define SHIFT_LOOP_HELPER(name, type, intrinsic, arg) \ + static AOM_INLINE void name(const type *in, type *out, int size) { \ + int i = 0; \ + do { \ + out[i] = intrinsic(in[i], arg); \ + } while (++i < size); \ + } + +SHIFT_LOOP_HELPER(shift_left_2_s16_x4, int16x4_t, vshl_n_s16, 2) +SHIFT_LOOP_HELPER(shift_left_2_s16_x8, int16x8_t, vshlq_n_s16, 2) +SHIFT_LOOP_HELPER(shift_left_2_s32_x4, int32x4_t, vshlq_n_s32, 2) +SHIFT_LOOP_HELPER(shift_right_2_round_s16_x8, int16x8_t, vrshrq_n_s16, 2) +SHIFT_LOOP_HELPER(shift_right_2_round_s32_x4, int32x4_t, vrshrq_n_s32, 2) +SHIFT_LOOP_HELPER(shift_right_4_round_s16_x8, int16x8_t, vrshrq_n_s16, 4) +SHIFT_LOOP_HELPER(shift_right_4_round_s32_x4, int32x4_t, vrshrq_n_s32, 4) + +// Addition instructions have slightly better performance compared to shift +// instructions on some micro-architectures, so use these for shifts by one. + +SHIFT_LOOP_HELPER(shift_left_1_s16_x4, int16x4_t, vadd_s16, in[i]) +SHIFT_LOOP_HELPER(shift_left_1_s16_x8, int16x8_t, vaddq_s16, in[i]) +SHIFT_LOOP_HELPER(shift_right_1_round_s16_x4, int16x4_t, vrhadd_s16, + vdup_n_s16(0)) +SHIFT_LOOP_HELPER(shift_right_1_round_s16_x8, int16x8_t, vrhaddq_s16, + vdupq_n_s16(0)) +SHIFT_LOOP_HELPER(shift_right_1_round_s32_x4, int32x4_t, vrhaddq_s32, + vdupq_n_s32(0)) + +#undef SHIFT_LOOP_HELPER + +#endif // AOM_AV1_ENCODER_ARM_NEON_SHIFT_NEON_H_ |