diff options
Diffstat (limited to 'third_party/aom/aom_dsp/simd/v256_intrinsics_x86.h')
-rw-r--r-- | third_party/aom/aom_dsp/simd/v256_intrinsics_x86.h | 754 |
1 files changed, 754 insertions, 0 deletions
diff --git a/third_party/aom/aom_dsp/simd/v256_intrinsics_x86.h b/third_party/aom/aom_dsp/simd/v256_intrinsics_x86.h new file mode 100644 index 0000000000..894ddee167 --- /dev/null +++ b/third_party/aom/aom_dsp/simd/v256_intrinsics_x86.h @@ -0,0 +1,754 @@ +/* + * Copyright (c) 2016, 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_AOM_DSP_SIMD_V256_INTRINSICS_X86_H_ +#define AOM_AOM_DSP_SIMD_V256_INTRINSICS_X86_H_ + +#if !defined(__AVX2__) + +#include "aom_dsp/simd/v256_intrinsics_v128.h" + +#else + +// The _m256i type seems to cause problems for g++'s mangling prior to +// version 5, but adding -fabi-version=0 fixes this. +#if !defined(__clang__) && defined(__GNUC__) && __GNUC__ < 5 && \ + defined(__AVX2__) && defined(__cplusplus) +#pragma GCC optimize "-fabi-version=0" +#endif + +#include <immintrin.h> + +#include "aom_dsp/simd/v128_intrinsics_x86.h" + +typedef __m256i v256; + +SIMD_INLINE uint32_t v256_low_u32(v256 a) { + return (uint32_t)_mm_cvtsi128_si32(_mm256_extracti128_si256(a, 0)); +} + +SIMD_INLINE v64 v256_low_v64(v256 a) { + return _mm_unpacklo_epi64(_mm256_extracti128_si256(a, 0), v64_zero()); +} + +SIMD_INLINE uint64_t v256_low_u64(v256 a) { return v64_u64(v256_low_v64(a)); } + +SIMD_INLINE v128 v256_low_v128(v256 a) { return _mm256_castsi256_si128(a); } + +SIMD_INLINE v128 v256_high_v128(v256 a) { + return _mm256_extracti128_si256(a, 1); +} + +SIMD_INLINE v256 v256_from_v128(v128 a, v128 b) { + // gcc seems to be missing _mm256_set_m128i() + return _mm256_inserti128_si256(_mm256_castsi128_si256(b), a, 1); +} + +SIMD_INLINE v256 v256_from_v64(v64 a, v64 b, v64 c, v64 d) { + return v256_from_v128(v128_from_v64(a, b), v128_from_v64(c, d)); +} + +SIMD_INLINE v256 v256_from_64(uint64_t a, uint64_t b, uint64_t c, uint64_t d) { + return _mm256_set_epi64x((int64_t)a, (int64_t)b, (int64_t)c, (int64_t)d); +} + +SIMD_INLINE v256 v256_load_aligned(const void *p) { + return _mm256_load_si256((const __m256i *)p); +} + +SIMD_INLINE v256 v256_load_unaligned(const void *p) { + return _mm256_loadu_si256((const __m256i *)p); +} + +SIMD_INLINE void v256_store_aligned(void *p, v256 a) { + _mm256_store_si256((__m256i *)p, a); +} + +SIMD_INLINE void v256_store_unaligned(void *p, v256 a) { + _mm256_storeu_si256((__m256i *)p, a); +} + +SIMD_INLINE v256 v256_zero(void) { return _mm256_setzero_si256(); } + +SIMD_INLINE v256 v256_dup_8(uint8_t x) { return _mm256_set1_epi8((char)x); } + +SIMD_INLINE v256 v256_dup_16(uint16_t x) { return _mm256_set1_epi16((short)x); } + +SIMD_INLINE v256 v256_dup_32(uint32_t x) { return _mm256_set1_epi32((int)x); } + +SIMD_INLINE v256 v256_dup_64(uint64_t x) { + return _mm256_set1_epi64x((int64_t)x); +} + +SIMD_INLINE v256 v256_add_8(v256 a, v256 b) { return _mm256_add_epi8(a, b); } + +SIMD_INLINE v256 v256_add_16(v256 a, v256 b) { return _mm256_add_epi16(a, b); } + +SIMD_INLINE v256 v256_sadd_u8(v256 a, v256 b) { return _mm256_adds_epu8(a, b); } + +SIMD_INLINE v256 v256_sadd_s8(v256 a, v256 b) { return _mm256_adds_epi8(a, b); } + +SIMD_INLINE v256 v256_sadd_s16(v256 a, v256 b) { + return _mm256_adds_epi16(a, b); +} + +SIMD_INLINE v256 v256_add_32(v256 a, v256 b) { return _mm256_add_epi32(a, b); } + +SIMD_INLINE v256 v256_add_64(v256 a, v256 b) { return _mm256_add_epi64(a, b); } + +SIMD_INLINE v256 v256_padd_u8(v256 a) { + return _mm256_maddubs_epi16(a, _mm256_set1_epi8(1)); +} + +SIMD_INLINE v256 v256_padd_s16(v256 a) { + return _mm256_madd_epi16(a, _mm256_set1_epi16(1)); +} + +SIMD_INLINE v256 v256_sub_8(v256 a, v256 b) { return _mm256_sub_epi8(a, b); } + +SIMD_INLINE v256 v256_ssub_u8(v256 a, v256 b) { return _mm256_subs_epu8(a, b); } + +SIMD_INLINE v256 v256_ssub_s8(v256 a, v256 b) { return _mm256_subs_epi8(a, b); } + +SIMD_INLINE v256 v256_sub_16(v256 a, v256 b) { return _mm256_sub_epi16(a, b); } + +SIMD_INLINE v256 v256_ssub_s16(v256 a, v256 b) { + return _mm256_subs_epi16(a, b); +} + +SIMD_INLINE v256 v256_ssub_u16(v256 a, v256 b) { + return _mm256_subs_epu16(a, b); +} + +SIMD_INLINE v256 v256_sub_32(v256 a, v256 b) { return _mm256_sub_epi32(a, b); } + +SIMD_INLINE v256 v256_sub_64(v256 a, v256 b) { return _mm256_sub_epi64(a, b); } + +SIMD_INLINE v256 v256_abs_s16(v256 a) { return _mm256_abs_epi16(a); } + +SIMD_INLINE v256 v256_abs_s8(v256 a) { return _mm256_abs_epi8(a); } + +// AVX doesn't have the direct intrinsics to zip/unzip 8, 16, 32 bit +// lanes of lower or upper halves of a 256bit vector because the +// unpack/pack intrinsics operate on the 256 bit input vector as 2 +// independent 128 bit vectors. +SIMD_INLINE v256 v256_ziplo_8(v256 a, v256 b) { + return _mm256_unpacklo_epi8( + _mm256_permute4x64_epi64(b, _MM_SHUFFLE(3, 1, 2, 0)), + _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0))); +} + +SIMD_INLINE v256 v256_ziphi_8(v256 a, v256 b) { + return _mm256_unpackhi_epi8( + _mm256_permute4x64_epi64(b, _MM_SHUFFLE(3, 1, 2, 0)), + _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0))); +} + +SIMD_INLINE v256 v256_ziplo_16(v256 a, v256 b) { + return _mm256_unpacklo_epi16( + _mm256_permute4x64_epi64(b, _MM_SHUFFLE(3, 1, 2, 0)), + _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0))); +} + +SIMD_INLINE v256 v256_ziphi_16(v256 a, v256 b) { + return _mm256_unpackhi_epi16( + _mm256_permute4x64_epi64(b, _MM_SHUFFLE(3, 1, 2, 0)), + _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0))); +} + +SIMD_INLINE v256 v256_ziplo_32(v256 a, v256 b) { + return _mm256_unpacklo_epi32( + _mm256_permute4x64_epi64(b, _MM_SHUFFLE(3, 1, 2, 0)), + _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0))); +} + +SIMD_INLINE v256 v256_ziphi_32(v256 a, v256 b) { + return _mm256_unpackhi_epi32( + _mm256_permute4x64_epi64(b, _MM_SHUFFLE(3, 1, 2, 0)), + _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0))); +} + +SIMD_INLINE v256 v256_ziplo_64(v256 a, v256 b) { + return _mm256_unpacklo_epi64( + _mm256_permute4x64_epi64(b, _MM_SHUFFLE(3, 1, 2, 0)), + _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0))); +} + +SIMD_INLINE v256 v256_ziphi_64(v256 a, v256 b) { + return _mm256_unpackhi_epi64( + _mm256_permute4x64_epi64(b, _MM_SHUFFLE(3, 1, 2, 0)), + _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0))); +} + +SIMD_INLINE v256 v256_ziplo_128(v256 a, v256 b) { + return _mm256_permute2x128_si256(a, b, 0x02); +} + +SIMD_INLINE v256 v256_ziphi_128(v256 a, v256 b) { + return _mm256_permute2x128_si256(a, b, 0x13); +} + +SIMD_INLINE v256 v256_zip_8(v128 a, v128 b) { + return v256_from_v128(v128_ziphi_8(a, b), v128_ziplo_8(a, b)); +} + +SIMD_INLINE v256 v256_zip_16(v128 a, v128 b) { + return v256_from_v128(v128_ziphi_16(a, b), v128_ziplo_16(a, b)); +} + +SIMD_INLINE v256 v256_zip_32(v128 a, v128 b) { + return v256_from_v128(v128_ziphi_32(a, b), v128_ziplo_32(a, b)); +} + +SIMD_INLINE v256 v256_unziphi_8(v256 a, v256 b) { + return _mm256_permute4x64_epi64( + _mm256_packs_epi16(_mm256_srai_epi16(b, 8), _mm256_srai_epi16(a, 8)), + _MM_SHUFFLE(3, 1, 2, 0)); +} + +SIMD_INLINE v256 v256_unziplo_8(v256 a, v256 b) { + return v256_unziphi_8(_mm256_slli_si256(a, 1), _mm256_slli_si256(b, 1)); +} + +SIMD_INLINE v256 v256_unziphi_16(v256 a, v256 b) { + return _mm256_permute4x64_epi64( + _mm256_packs_epi32(_mm256_srai_epi32(b, 16), _mm256_srai_epi32(a, 16)), + _MM_SHUFFLE(3, 1, 2, 0)); +} + +SIMD_INLINE v256 v256_unziplo_16(v256 a, v256 b) { + return v256_unziphi_16(_mm256_slli_si256(a, 2), _mm256_slli_si256(b, 2)); +} + +SIMD_INLINE v256 v256_unziphi_32(v256 a, v256 b) { + return _mm256_permute4x64_epi64( + _mm256_castps_si256(_mm256_shuffle_ps(_mm256_castsi256_ps(b), + _mm256_castsi256_ps(a), + _MM_SHUFFLE(3, 1, 3, 1))), + _MM_SHUFFLE(3, 1, 2, 0)); +} + +SIMD_INLINE v256 v256_unziplo_32(v256 a, v256 b) { + return _mm256_permute4x64_epi64( + _mm256_castps_si256(_mm256_shuffle_ps(_mm256_castsi256_ps(b), + _mm256_castsi256_ps(a), + _MM_SHUFFLE(2, 0, 2, 0))), + _MM_SHUFFLE(3, 1, 2, 0)); +} + +SIMD_INLINE v256 v256_unziphi_64(v256 a, v256 b) { + return _mm256_permute4x64_epi64( + _mm256_castpd_si256(_mm256_shuffle_pd(_mm256_castsi256_pd(b), + _mm256_castsi256_pd(a), 15)), + _MM_SHUFFLE(3, 1, 2, 0)); +} + +SIMD_INLINE v256 v256_unziplo_64(v256 a, v256 b) { + return _mm256_permute4x64_epi64( + _mm256_castpd_si256( + _mm256_shuffle_pd(_mm256_castsi256_pd(b), _mm256_castsi256_pd(a), 0)), + _MM_SHUFFLE(3, 1, 2, 0)); +} + +SIMD_INLINE v256 v256_unpack_u8_s16(v128 a) { return _mm256_cvtepu8_epi16(a); } + +SIMD_INLINE v256 v256_unpacklo_u8_s16(v256 a) { + return _mm256_unpacklo_epi8( + _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0)), + _mm256_setzero_si256()); +} + +SIMD_INLINE v256 v256_unpackhi_u8_s16(v256 a) { + return _mm256_unpackhi_epi8( + _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0)), + _mm256_setzero_si256()); +} + +SIMD_INLINE v256 v256_unpack_s8_s16(v128 a) { + return v256_from_v128(v128_unpackhi_s8_s16(a), v128_unpacklo_s8_s16(a)); +} + +SIMD_INLINE v256 v256_unpacklo_s8_s16(v256 a) { + return _mm256_srai_epi16( + _mm256_unpacklo_epi8( + a, _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0))), + 8); +} + +SIMD_INLINE v256 v256_unpackhi_s8_s16(v256 a) { + return _mm256_srai_epi16( + _mm256_unpackhi_epi8( + a, _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0))), + 8); +} + +SIMD_INLINE v256 v256_pack_s32_s16(v256 a, v256 b) { + return _mm256_permute4x64_epi64(_mm256_packs_epi32(b, a), + _MM_SHUFFLE(3, 1, 2, 0)); +} + +SIMD_INLINE v256 v256_pack_s32_u16(v256 a, v256 b) { + return _mm256_permute4x64_epi64(_mm256_packus_epi32(b, a), + _MM_SHUFFLE(3, 1, 2, 0)); +} + +SIMD_INLINE v256 v256_pack_s16_u8(v256 a, v256 b) { + return _mm256_permute4x64_epi64(_mm256_packus_epi16(b, a), + _MM_SHUFFLE(3, 1, 2, 0)); +} + +SIMD_INLINE v256 v256_pack_s16_s8(v256 a, v256 b) { + return _mm256_permute4x64_epi64(_mm256_packs_epi16(b, a), + _MM_SHUFFLE(3, 1, 2, 0)); +} + +SIMD_INLINE v256 v256_unpack_u16_s32(v128 a) { + return _mm256_cvtepu16_epi32(a); +} + +SIMD_INLINE v256 v256_unpack_s16_s32(v128 a) { + return _mm256_cvtepi16_epi32(a); +} + +SIMD_INLINE v256 v256_unpacklo_u16_s32(v256 a) { + return _mm256_unpacklo_epi16( + _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0)), + _mm256_setzero_si256()); +} + +SIMD_INLINE v256 v256_unpacklo_s16_s32(v256 a) { + return _mm256_srai_epi32( + _mm256_unpacklo_epi16( + a, _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0))), + 16); +} + +SIMD_INLINE v256 v256_unpackhi_u16_s32(v256 a) { + return _mm256_unpackhi_epi16( + _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0)), + _mm256_setzero_si256()); +} + +SIMD_INLINE v256 v256_unpackhi_s16_s32(v256 a) { + return _mm256_srai_epi32( + _mm256_unpackhi_epi16( + a, _mm256_permute4x64_epi64(a, _MM_SHUFFLE(3, 1, 2, 0))), + 16); +} + +SIMD_INLINE v256 v256_shuffle_8(v256 a, v256 pattern) { + return _mm256_blendv_epi8( + _mm256_shuffle_epi8( + _mm256_permute2x128_si256(a, a, _MM_SHUFFLE(0, 1, 0, 1)), pattern), + _mm256_shuffle_epi8( + _mm256_permute2x128_si256(a, a, _MM_SHUFFLE(0, 0, 0, 0)), pattern), + _mm256_cmpgt_epi8(v256_dup_8(16), pattern)); +} + +SIMD_INLINE v256 v256_wideshuffle_8(v256 a, v256 b, v256 pattern) { + v256 c32 = v256_dup_8(32); + v256 p32 = v256_sub_8(pattern, c32); + v256 r1 = _mm256_blendv_epi8( + _mm256_shuffle_epi8( + _mm256_permute2x128_si256(a, b, _MM_SHUFFLE(0, 1, 0, 1)), p32), + _mm256_shuffle_epi8( + _mm256_permute2x128_si256(a, b, _MM_SHUFFLE(0, 0, 0, 0)), p32), + _mm256_cmpgt_epi8(v256_dup_8(48), pattern)); + v256 r2 = _mm256_blendv_epi8( + _mm256_shuffle_epi8( + _mm256_permute2x128_si256(a, b, _MM_SHUFFLE(0, 3, 0, 3)), pattern), + _mm256_shuffle_epi8( + _mm256_permute2x128_si256(a, b, _MM_SHUFFLE(0, 2, 0, 2)), pattern), + _mm256_cmpgt_epi8(v256_dup_8(16), pattern)); + return _mm256_blendv_epi8(r1, r2, _mm256_cmpgt_epi8(c32, pattern)); +} + +SIMD_INLINE v256 v256_pshuffle_8(v256 a, v256 pattern) { + return _mm256_shuffle_epi8(a, pattern); +} + +SIMD_INLINE int64_t v256_dotp_su8(v256 a, v256 b) { + v256 t1 = _mm256_madd_epi16(v256_unpackhi_s8_s16(a), v256_unpackhi_u8_s16(b)); + v256 t2 = _mm256_madd_epi16(v256_unpacklo_s8_s16(a), v256_unpacklo_u8_s16(b)); + t1 = _mm256_add_epi32(t1, t2); + v128 t = _mm_add_epi32(_mm256_extracti128_si256(t1, 0), + _mm256_extracti128_si256(t1, 1)); + t = _mm_add_epi32(t, _mm_srli_si128(t, 8)); + t = _mm_add_epi32(t, _mm_srli_si128(t, 4)); + return (int32_t)v128_low_u32(t); +} + +SIMD_INLINE int64_t v256_dotp_s16(v256 a, v256 b) { + v256 r = _mm256_madd_epi16(a, b); +#if defined(__x86_64__) + v128 t; + r = _mm256_add_epi64(_mm256_cvtepi32_epi64(v256_high_v128(r)), + _mm256_cvtepi32_epi64(v256_low_v128(r))); + t = v256_low_v128(_mm256_add_epi64( + r, _mm256_permute2x128_si256(r, r, _MM_SHUFFLE(2, 0, 0, 1)))); + return _mm_cvtsi128_si64(_mm_add_epi64(t, _mm_srli_si128(t, 8))); +#else + v128 l = v256_low_v128(r); + v128 h = v256_high_v128(r); + return (int64_t)_mm_cvtsi128_si32(l) + + (int64_t)_mm_cvtsi128_si32(_mm_srli_si128(l, 4)) + + (int64_t)_mm_cvtsi128_si32(_mm_srli_si128(l, 8)) + + (int64_t)_mm_cvtsi128_si32(_mm_srli_si128(l, 12)) + + (int64_t)_mm_cvtsi128_si32(h) + + (int64_t)_mm_cvtsi128_si32(_mm_srli_si128(h, 4)) + + (int64_t)_mm_cvtsi128_si32(_mm_srli_si128(h, 8)) + + (int64_t)_mm_cvtsi128_si32(_mm_srli_si128(h, 12)); +#endif +} + +SIMD_INLINE int64_t v256_dotp_s32(v256 a, v256 b) { + v256 r = _mm256_mullo_epi32(a, b); +#if defined(__x86_64__) + v128 t; + r = _mm256_add_epi64(_mm256_cvtepi32_epi64(v256_high_v128(r)), + _mm256_cvtepi32_epi64(v256_low_v128(r))); + t = v256_low_v128(_mm256_add_epi64( + r, _mm256_permute2x128_si256(r, r, _MM_SHUFFLE(2, 0, 0, 1)))); + return _mm_cvtsi128_si64(_mm_add_epi64(t, _mm_srli_si128(t, 8))); +#else + v128 l = v256_low_v128(r); + v128 h = v256_high_v128(r); + return (int64_t)_mm_cvtsi128_si32(l) + + (int64_t)_mm_cvtsi128_si32(_mm_srli_si128(l, 4)) + + (int64_t)_mm_cvtsi128_si32(_mm_srli_si128(l, 8)) + + (int64_t)_mm_cvtsi128_si32(_mm_srli_si128(l, 12)) + + (int64_t)_mm_cvtsi128_si32(h) + + (int64_t)_mm_cvtsi128_si32(_mm_srli_si128(h, 4)) + + (int64_t)_mm_cvtsi128_si32(_mm_srli_si128(h, 8)) + + (int64_t)_mm_cvtsi128_si32(_mm_srli_si128(h, 12)); +#endif +} + +SIMD_INLINE uint64_t v256_hadd_u8(v256 a) { + v256 t = _mm256_sad_epu8(a, _mm256_setzero_si256()); + v128 lo = v256_low_v128(t); + v128 hi = v256_high_v128(t); + lo = v128_add_32(lo, hi); + return v64_low_u32(v128_low_v64(lo)) + v128_low_u32(v128_high_v64(lo)); +} + +typedef v256 sad256_internal; + +SIMD_INLINE sad256_internal v256_sad_u8_init(void) { + return _mm256_setzero_si256(); +} + +/* Implementation dependent return value. Result must be finalised with + v256_sad_u8_sum(). + The result for more than 32 v256_sad_u8() calls is undefined. */ +SIMD_INLINE sad256_internal v256_sad_u8(sad256_internal s, v256 a, v256 b) { + return _mm256_add_epi64(s, _mm256_sad_epu8(a, b)); +} + +SIMD_INLINE uint32_t v256_sad_u8_sum(sad256_internal s) { + v256 t = _mm256_add_epi32(s, _mm256_unpackhi_epi64(s, s)); + return v128_low_u32(_mm_add_epi32(v256_high_v128(t), v256_low_v128(t))); +} + +typedef v256 ssd256_internal; + +SIMD_INLINE ssd256_internal v256_ssd_u8_init(void) { + return _mm256_setzero_si256(); +} + +/* Implementation dependent return value. Result must be finalised with + * v256_ssd_u8_sum(). */ +SIMD_INLINE ssd256_internal v256_ssd_u8(ssd256_internal s, v256 a, v256 b) { + v256 l = _mm256_sub_epi16(_mm256_unpacklo_epi8(a, _mm256_setzero_si256()), + _mm256_unpacklo_epi8(b, _mm256_setzero_si256())); + v256 h = _mm256_sub_epi16(_mm256_unpackhi_epi8(a, _mm256_setzero_si256()), + _mm256_unpackhi_epi8(b, _mm256_setzero_si256())); + v256 rl = _mm256_madd_epi16(l, l); + v256 rh = _mm256_madd_epi16(h, h); + v128 c = _mm_cvtsi32_si128(32); + rl = _mm256_add_epi32(rl, _mm256_srli_si256(rl, 8)); + rl = _mm256_add_epi32(rl, _mm256_srli_si256(rl, 4)); + rh = _mm256_add_epi32(rh, _mm256_srli_si256(rh, 8)); + rh = _mm256_add_epi32(rh, _mm256_srli_si256(rh, 4)); + return _mm256_add_epi64( + s, + _mm256_srl_epi64(_mm256_sll_epi64(_mm256_unpacklo_epi64(rl, rh), c), c)); +} + +SIMD_INLINE uint32_t v256_ssd_u8_sum(ssd256_internal s) { + v256 t = _mm256_add_epi32(s, _mm256_unpackhi_epi64(s, s)); + return v128_low_u32(_mm_add_epi32(v256_high_v128(t), v256_low_v128(t))); +} + +SIMD_INLINE v256 v256_or(v256 a, v256 b) { return _mm256_or_si256(a, b); } + +SIMD_INLINE v256 v256_xor(v256 a, v256 b) { return _mm256_xor_si256(a, b); } + +SIMD_INLINE v256 v256_and(v256 a, v256 b) { return _mm256_and_si256(a, b); } + +SIMD_INLINE v256 v256_andn(v256 a, v256 b) { return _mm256_andnot_si256(b, a); } + +SIMD_INLINE v256 v256_mul_s16(v64 a, v64 b) { + v128 lo_bits = v128_mullo_s16(a, b); + v128 hi_bits = v128_mulhi_s16(a, b); + return v256_from_v128(v128_ziphi_16(hi_bits, lo_bits), + v128_ziplo_16(hi_bits, lo_bits)); +} + +SIMD_INLINE v256 v256_mullo_s16(v256 a, v256 b) { + return _mm256_mullo_epi16(a, b); +} + +SIMD_INLINE v256 v256_mulhi_s16(v256 a, v256 b) { + return _mm256_mulhi_epi16(a, b); +} + +SIMD_INLINE v256 v256_mullo_s32(v256 a, v256 b) { + return _mm256_mullo_epi32(a, b); +} + +SIMD_INLINE v256 v256_madd_s16(v256 a, v256 b) { + return _mm256_madd_epi16(a, b); +} + +SIMD_INLINE v256 v256_madd_us8(v256 a, v256 b) { + return _mm256_maddubs_epi16(a, b); +} + +SIMD_INLINE v256 v256_avg_u8(v256 a, v256 b) { return _mm256_avg_epu8(a, b); } + +SIMD_INLINE v256 v256_rdavg_u8(v256 a, v256 b) { + return _mm256_sub_epi8( + _mm256_avg_epu8(a, b), + _mm256_and_si256(_mm256_xor_si256(a, b), v256_dup_8(1))); +} + +SIMD_INLINE v256 v256_rdavg_u16(v256 a, v256 b) { + return _mm256_sub_epi16( + _mm256_avg_epu16(a, b), + _mm256_and_si256(_mm256_xor_si256(a, b), v256_dup_16(1))); +} + +SIMD_INLINE v256 v256_avg_u16(v256 a, v256 b) { return _mm256_avg_epu16(a, b); } + +SIMD_INLINE v256 v256_min_u8(v256 a, v256 b) { return _mm256_min_epu8(a, b); } + +SIMD_INLINE v256 v256_max_u8(v256 a, v256 b) { return _mm256_max_epu8(a, b); } + +SIMD_INLINE v256 v256_min_s8(v256 a, v256 b) { return _mm256_min_epi8(a, b); } + +SIMD_INLINE uint32_t v256_movemask_8(v256 a) { + return (uint32_t)_mm256_movemask_epi8(a); +} + +SIMD_INLINE v256 v256_blend_8(v256 a, v256 b, v256 c) { + return _mm256_blendv_epi8(a, b, c); +} + +SIMD_INLINE v256 v256_max_s8(v256 a, v256 b) { return _mm256_max_epi8(a, b); } + +SIMD_INLINE v256 v256_min_s16(v256 a, v256 b) { return _mm256_min_epi16(a, b); } + +SIMD_INLINE v256 v256_max_s16(v256 a, v256 b) { return _mm256_max_epi16(a, b); } + +SIMD_INLINE v256 v256_min_s32(v256 a, v256 b) { return _mm256_min_epi32(a, b); } + +SIMD_INLINE v256 v256_max_s32(v256 a, v256 b) { return _mm256_max_epi32(a, b); } + +SIMD_INLINE v256 v256_cmpgt_s8(v256 a, v256 b) { + return _mm256_cmpgt_epi8(a, b); +} + +SIMD_INLINE v256 v256_cmplt_s8(v256 a, v256 b) { + return _mm256_cmpgt_epi8(b, a); +} + +SIMD_INLINE v256 v256_cmpeq_8(v256 a, v256 b) { + return _mm256_cmpeq_epi8(a, b); +} + +SIMD_INLINE v256 v256_cmpgt_s16(v256 a, v256 b) { + return _mm256_cmpgt_epi16(a, b); +} + +SIMD_INLINE v256 v256_cmplt_s16(v256 a, v256 b) { + return _mm256_cmpgt_epi16(b, a); +} + +SIMD_INLINE v256 v256_cmpeq_16(v256 a, v256 b) { + return _mm256_cmpeq_epi16(a, b); +} + +SIMD_INLINE v256 v256_cmpgt_s32(v256 a, v256 b) { + return _mm256_cmpgt_epi32(a, b); +} + +SIMD_INLINE v256 v256_cmplt_s32(v256 a, v256 b) { + return _mm256_cmpgt_epi32(b, a); +} + +SIMD_INLINE v256 v256_cmpeq_32(v256 a, v256 b) { + return _mm256_cmpeq_epi32(a, b); +} + +SIMD_INLINE v256 v256_shl_8(v256 a, unsigned int c) { + return _mm256_and_si256(_mm256_set1_epi8((char)(0xff << c)), + _mm256_sll_epi16(a, _mm_cvtsi32_si128((int)c))); +} + +SIMD_INLINE v256 v256_shr_u8(v256 a, unsigned int c) { + return _mm256_and_si256(_mm256_set1_epi8((char)(0xff >> c)), + _mm256_srl_epi16(a, _mm_cvtsi32_si128((int)c))); +} + +SIMD_INLINE v256 v256_shr_s8(v256 a, unsigned int c) { + __m128i x = _mm_cvtsi32_si128((int)(c + 8)); + return _mm256_packs_epi16(_mm256_sra_epi16(_mm256_unpacklo_epi8(a, a), x), + _mm256_sra_epi16(_mm256_unpackhi_epi8(a, a), x)); +} + +SIMD_INLINE v256 v256_shl_16(v256 a, unsigned int c) { + return _mm256_sll_epi16(a, _mm_cvtsi32_si128((int)c)); +} + +SIMD_INLINE v256 v256_shr_u16(v256 a, unsigned int c) { + return _mm256_srl_epi16(a, _mm_cvtsi32_si128((int)c)); +} + +SIMD_INLINE v256 v256_shr_s16(v256 a, unsigned int c) { + return _mm256_sra_epi16(a, _mm_cvtsi32_si128((int)c)); +} + +SIMD_INLINE v256 v256_shl_32(v256 a, unsigned int c) { + return _mm256_sll_epi32(a, _mm_cvtsi32_si128((int)c)); +} + +SIMD_INLINE v256 v256_shr_u32(v256 a, unsigned int c) { + return _mm256_srl_epi32(a, _mm_cvtsi32_si128((int)c)); +} + +SIMD_INLINE v256 v256_shr_s32(v256 a, unsigned int c) { + return _mm256_sra_epi32(a, _mm_cvtsi32_si128((int)c)); +} + +SIMD_INLINE v256 v256_shl_64(v256 a, unsigned int c) { + return _mm256_sll_epi64(a, _mm_cvtsi32_si128((int)c)); +} + +SIMD_INLINE v256 v256_shr_u64(v256 a, unsigned int c) { + return _mm256_srl_epi64(a, _mm_cvtsi32_si128((int)c)); +} + +SIMD_INLINE v256 v256_shr_s64(v256 a, unsigned int c) { +#if defined(__AVX512VL__) + return _mm256_sra_epi64(a, _mm_cvtsi32_si128((int)c)); +#else + return v256_from_v128(v128_shr_s64(v256_high_v128(a), c), + v128_shr_s64(v256_low_v128(a), c)); +#endif +} + +/* These intrinsics require immediate values, so we must use #defines + to enforce that. */ +// _mm256_slli_si256 works on 128 bit lanes and can't be used +#define v256_shl_n_byte(a, n) \ + ((n) < 16 ? v256_from_v128( \ + v128_align(v256_high_v128(a), v256_low_v128(a), 16 - (n)), \ + v128_shl_n_byte(v256_low_v128(a), n)) \ + : _mm256_inserti128_si256( \ + _mm256_setzero_si256(), \ + v128_shl_n_byte(v256_low_v128(a), (n)-16), 1)) + +// _mm256_srli_si256 works on 128 bit lanes and can't be used +#define v256_shr_n_byte(a, n) \ + ((n) < 16 \ + ? _mm256_alignr_epi8( \ + _mm256_permute2x128_si256(a, a, _MM_SHUFFLE(2, 0, 0, 1)), a, n) \ + : ((n) == 16 ? _mm256_permute2x128_si256(_mm256_setzero_si256(), a, 3) \ + : _mm256_inserti128_si256( \ + _mm256_setzero_si256(), \ + v128_shr_n_byte(v256_high_v128(a), (n)-16), 0))) + +// _mm256_alignr_epi8 works on two 128 bit lanes and can't be used +#define v256_align(a, b, c) \ + ((c) ? v256_or(v256_shr_n_byte(b, c), v256_shl_n_byte(a, 32 - (c))) : b) + +#define v256_shl_n_8(a, c) \ + _mm256_and_si256(_mm256_set1_epi8((char)(0xff << (c))), \ + _mm256_slli_epi16(a, c)) +#define v256_shr_n_u8(a, c) \ + _mm256_and_si256(_mm256_set1_epi8((char)(0xff >> (c))), \ + _mm256_srli_epi16(a, c)) +#define v256_shr_n_s8(a, c) \ + _mm256_packs_epi16(_mm256_srai_epi16(_mm256_unpacklo_epi8(a, a), (c) + 8), \ + _mm256_srai_epi16(_mm256_unpackhi_epi8(a, a), (c) + 8)) +#define v256_shl_n_16(a, c) _mm256_slli_epi16(a, c) +#define v256_shr_n_u16(a, c) _mm256_srli_epi16(a, c) +#define v256_shr_n_s16(a, c) _mm256_srai_epi16(a, c) +#define v256_shl_n_32(a, c) _mm256_slli_epi32(a, c) +#define v256_shr_n_u32(a, c) _mm256_srli_epi32(a, c) +#define v256_shr_n_s32(a, c) _mm256_srai_epi32(a, c) +#define v256_shl_n_64(a, c) _mm256_slli_epi64(a, c) +#define v256_shr_n_u64(a, c) _mm256_srli_epi64(a, c) +#define v256_shr_n_s64(a, c) \ + v256_shr_s64((a), (c)) // _mm256_srai_epi64 broken in gcc? +#define v256_shr_n_word(a, n) v256_shr_n_byte(a, 2 * (n)) +#define v256_shl_n_word(a, n) v256_shl_n_byte(a, 2 * (n)) + +typedef v256 sad256_internal_u16; + +SIMD_INLINE sad256_internal_u16 v256_sad_u16_init(void) { return v256_zero(); } + +/* Implementation dependent return value. Result must be finalised with + * v256_sad_u16_sum(). */ +SIMD_INLINE sad256_internal_u16 v256_sad_u16(sad256_internal_u16 s, v256 a, + v256 b) { +#if defined(__SSE4_1__) + v256 t = v256_sub_16(_mm256_max_epu16(a, b), _mm256_min_epu16(a, b)); +#else + v256 t = v256_cmplt_s16(v256_xor(a, v256_dup_16(32768)), + v256_xor(b, v256_dup_16(32768))); + t = v256_sub_16(v256_or(v256_and(b, t), v256_andn(a, t)), + v256_or(v256_and(a, t), v256_andn(b, t))); +#endif + return v256_add_32( + s, v256_add_32(v256_unpackhi_u16_s32(t), v256_unpacklo_u16_s32(t))); +} + +SIMD_INLINE uint32_t v256_sad_u16_sum(sad256_internal_u16 s) { + v128 t = v128_add_32(v256_high_v128(s), v256_low_v128(s)); + return v128_low_u32(t) + v128_low_u32(v128_shr_n_byte(t, 4)) + + v128_low_u32(v128_shr_n_byte(t, 8)) + + v128_low_u32(v128_shr_n_byte(t, 12)); +} + +typedef v256 ssd256_internal_s16; + +SIMD_INLINE ssd256_internal_s16 v256_ssd_s16_init(void) { return v256_zero(); } + +/* Implementation dependent return value. Result must be finalised with + * v256_ssd_s16_sum(). */ +SIMD_INLINE ssd256_internal_s16 v256_ssd_s16(ssd256_internal_s16 s, v256 a, + v256 b) { + v256 d = v256_sub_16(a, b); + d = v256_madd_s16(d, d); + return v256_add_64(s, v256_add_64(_mm256_unpackhi_epi32(d, v256_zero()), + _mm256_unpacklo_epi32(d, v256_zero()))); +} + +SIMD_INLINE uint64_t v256_ssd_s16_sum(ssd256_internal_s16 s) { + v128 t = v128_add_64(v256_high_v128(s), v256_low_v128(s)); + return v64_u64(v128_low_v64(t)) + v64_u64(v128_high_v64(t)); +} + +#endif + +#endif // AOM_AOM_DSP_SIMD_V256_INTRINSICS_X86_H_ |