diff options
Diffstat (limited to '')
-rw-r--r-- | third_party/aom/aom_dsp/x86/quantize_avx2.c | 274 |
1 files changed, 274 insertions, 0 deletions
diff --git a/third_party/aom/aom_dsp/x86/quantize_avx2.c b/third_party/aom/aom_dsp/x86/quantize_avx2.c new file mode 100644 index 0000000000..b808d46778 --- /dev/null +++ b/third_party/aom/aom_dsp/x86/quantize_avx2.c @@ -0,0 +1,274 @@ +/* + * Copyright (c) 2022, 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. + */ + +#include <immintrin.h> +#include "config/aom_dsp_rtcd.h" +#include "aom/aom_integer.h" +#include "aom_dsp/x86/quantize_x86.h" + +static INLINE void load_b_values_avx2(const int16_t *zbin_ptr, __m256i *zbin, + const int16_t *round_ptr, __m256i *round, + const int16_t *quant_ptr, __m256i *quant, + const int16_t *dequant_ptr, + __m256i *dequant, + const int16_t *shift_ptr, __m256i *shift, + int log_scale) { + *zbin = _mm256_castsi128_si256(_mm_load_si128((const __m128i *)zbin_ptr)); + *zbin = _mm256_permute4x64_epi64(*zbin, 0x54); + if (log_scale > 0) { + const __m256i rnd = _mm256_set1_epi16((int16_t)(1 << (log_scale - 1))); + *zbin = _mm256_add_epi16(*zbin, rnd); + *zbin = _mm256_srai_epi16(*zbin, log_scale); + } + // Subtracting 1 here eliminates a _mm256_cmpeq_epi16() instruction when + // calculating the zbin mask. (See quantize_b_logscale{0,1,2}_16) + *zbin = _mm256_sub_epi16(*zbin, _mm256_set1_epi16(1)); + + *round = _mm256_castsi128_si256(_mm_load_si128((const __m128i *)round_ptr)); + *round = _mm256_permute4x64_epi64(*round, 0x54); + if (log_scale > 0) { + const __m256i rnd = _mm256_set1_epi16((int16_t)(1 << (log_scale - 1))); + *round = _mm256_add_epi16(*round, rnd); + *round = _mm256_srai_epi16(*round, log_scale); + } + + *quant = _mm256_castsi128_si256(_mm_load_si128((const __m128i *)quant_ptr)); + *quant = _mm256_permute4x64_epi64(*quant, 0x54); + *dequant = + _mm256_castsi128_si256(_mm_load_si128((const __m128i *)dequant_ptr)); + *dequant = _mm256_permute4x64_epi64(*dequant, 0x54); + *shift = _mm256_castsi128_si256(_mm_load_si128((const __m128i *)shift_ptr)); + *shift = _mm256_permute4x64_epi64(*shift, 0x54); +} + +static INLINE __m256i load_coefficients_avx2(const tran_low_t *coeff_ptr) { + const __m256i coeff1 = _mm256_load_si256((__m256i *)coeff_ptr); + const __m256i coeff2 = _mm256_load_si256((__m256i *)(coeff_ptr + 8)); + return _mm256_packs_epi32(coeff1, coeff2); +} + +static INLINE void store_coefficients_avx2(__m256i coeff_vals, + tran_low_t *coeff_ptr) { + __m256i coeff_sign = _mm256_srai_epi16(coeff_vals, 15); + __m256i coeff_vals_lo = _mm256_unpacklo_epi16(coeff_vals, coeff_sign); + __m256i coeff_vals_hi = _mm256_unpackhi_epi16(coeff_vals, coeff_sign); + _mm256_store_si256((__m256i *)coeff_ptr, coeff_vals_lo); + _mm256_store_si256((__m256i *)(coeff_ptr + 8), coeff_vals_hi); +} + +static AOM_FORCE_INLINE __m256i quantize_b_logscale0_16( + const tran_low_t *coeff_ptr, tran_low_t *qcoeff_ptr, + tran_low_t *dqcoeff_ptr, __m256i *v_quant, __m256i *v_dequant, + __m256i *v_round, __m256i *v_zbin, __m256i *v_quant_shift) { + const __m256i v_coeff = load_coefficients_avx2(coeff_ptr); + const __m256i v_abs_coeff = _mm256_abs_epi16(v_coeff); + const __m256i v_zbin_mask = _mm256_cmpgt_epi16(v_abs_coeff, *v_zbin); + + if (_mm256_movemask_epi8(v_zbin_mask) == 0) { + _mm256_store_si256((__m256i *)qcoeff_ptr, _mm256_setzero_si256()); + _mm256_store_si256((__m256i *)dqcoeff_ptr, _mm256_setzero_si256()); + _mm256_store_si256((__m256i *)(qcoeff_ptr + 8), _mm256_setzero_si256()); + _mm256_store_si256((__m256i *)(dqcoeff_ptr + 8), _mm256_setzero_si256()); + return _mm256_setzero_si256(); + } + + // tmp = v_zbin_mask ? (int64_t)abs_coeff + log_scaled_round : 0 + const __m256i v_tmp_rnd = + _mm256_and_si256(_mm256_adds_epi16(v_abs_coeff, *v_round), v_zbin_mask); + // tmp32 = (int)(((((tmp * quant_ptr[rc != 0]) >> 16) + tmp) * + // quant_shift_ptr[rc != 0]) >> + // (16 - log_scale + AOM_QM_BITS)); + const __m256i v_tmp32_a = _mm256_mulhi_epi16(v_tmp_rnd, *v_quant); + const __m256i v_tmp32_b = _mm256_add_epi16(v_tmp32_a, v_tmp_rnd); + const __m256i v_tmp32 = _mm256_mulhi_epi16(v_tmp32_b, *v_quant_shift); + const __m256i v_nz_mask = _mm256_cmpgt_epi16(v_tmp32, _mm256_setzero_si256()); + const __m256i v_qcoeff = _mm256_sign_epi16(v_tmp32, v_coeff); + const __m256i v_dqcoeff = _mm256_mullo_epi16(v_qcoeff, *v_dequant); + store_coefficients_avx2(v_qcoeff, qcoeff_ptr); + store_coefficients_avx2(v_dqcoeff, dqcoeff_ptr); + return v_nz_mask; +} + +static INLINE __m256i get_max_lane_eob(const int16_t *iscan, __m256i v_eobmax, + __m256i v_mask) { + const __m256i v_iscan = _mm256_loadu_si256((const __m256i *)iscan); + const __m256i v_iscan_perm = _mm256_permute4x64_epi64(v_iscan, 0xD8); + const __m256i v_iscan_plus1 = _mm256_sub_epi16(v_iscan_perm, v_mask); + const __m256i v_nz_iscan = _mm256_and_si256(v_iscan_plus1, v_mask); + return _mm256_max_epi16(v_eobmax, v_nz_iscan); +} + +static INLINE int16_t accumulate_eob256(__m256i eob256) { + const __m128i eob_lo = _mm256_castsi256_si128(eob256); + const __m128i eob_hi = _mm256_extractf128_si256(eob256, 1); + __m128i eob = _mm_max_epi16(eob_lo, eob_hi); + __m128i eob_shuffled = _mm_shuffle_epi32(eob, 0xe); + eob = _mm_max_epi16(eob, eob_shuffled); + eob_shuffled = _mm_shufflelo_epi16(eob, 0xe); + eob = _mm_max_epi16(eob, eob_shuffled); + eob_shuffled = _mm_shufflelo_epi16(eob, 0x1); + eob = _mm_max_epi16(eob, eob_shuffled); + return _mm_extract_epi16(eob, 1); +} + +void aom_quantize_b_avx2(const tran_low_t *coeff_ptr, intptr_t n_coeffs, + const int16_t *zbin_ptr, const int16_t *round_ptr, + const int16_t *quant_ptr, + const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, + tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, + uint16_t *eob_ptr, const int16_t *scan, + const int16_t *iscan) { + (void)scan; + __m256i v_zbin, v_round, v_quant, v_dequant, v_quant_shift; + __m256i v_eobmax = _mm256_setzero_si256(); + + load_b_values_avx2(zbin_ptr, &v_zbin, round_ptr, &v_round, quant_ptr, + &v_quant, dequant_ptr, &v_dequant, quant_shift_ptr, + &v_quant_shift, 0); + + // Do DC and first 15 AC. + __m256i v_nz_mask = + quantize_b_logscale0_16(coeff_ptr, qcoeff_ptr, dqcoeff_ptr, &v_quant, + &v_dequant, &v_round, &v_zbin, &v_quant_shift); + + v_eobmax = get_max_lane_eob(iscan, v_eobmax, v_nz_mask); + + v_round = _mm256_unpackhi_epi64(v_round, v_round); + v_quant = _mm256_unpackhi_epi64(v_quant, v_quant); + v_dequant = _mm256_unpackhi_epi64(v_dequant, v_dequant); + v_quant_shift = _mm256_unpackhi_epi64(v_quant_shift, v_quant_shift); + v_zbin = _mm256_unpackhi_epi64(v_zbin, v_zbin); + + for (intptr_t count = n_coeffs - 16; count > 0; count -= 16) { + coeff_ptr += 16; + qcoeff_ptr += 16; + dqcoeff_ptr += 16; + iscan += 16; + v_nz_mask = + quantize_b_logscale0_16(coeff_ptr, qcoeff_ptr, dqcoeff_ptr, &v_quant, + &v_dequant, &v_round, &v_zbin, &v_quant_shift); + + v_eobmax = get_max_lane_eob(iscan, v_eobmax, v_nz_mask); + } + + *eob_ptr = accumulate_eob256(v_eobmax); +} + +static AOM_FORCE_INLINE __m256i quantize_b_logscale_16( + const tran_low_t *coeff_ptr, tran_low_t *qcoeff_ptr, + tran_low_t *dqcoeff_ptr, __m256i *v_quant, __m256i *v_dequant, + __m256i *v_round, __m256i *v_zbin, __m256i *v_quant_shift, int log_scale) { + const __m256i v_coeff = load_coefficients_avx2(coeff_ptr); + const __m256i v_abs_coeff = _mm256_abs_epi16(v_coeff); + const __m256i v_zbin_mask = _mm256_cmpgt_epi16(v_abs_coeff, *v_zbin); + + if (_mm256_movemask_epi8(v_zbin_mask) == 0) { + _mm256_store_si256((__m256i *)qcoeff_ptr, _mm256_setzero_si256()); + _mm256_store_si256((__m256i *)dqcoeff_ptr, _mm256_setzero_si256()); + _mm256_store_si256((__m256i *)(qcoeff_ptr + 8), _mm256_setzero_si256()); + _mm256_store_si256((__m256i *)(dqcoeff_ptr + 8), _mm256_setzero_si256()); + return _mm256_setzero_si256(); + } + + // tmp = v_zbin_mask ? (int64_t)abs_coeff + log_scaled_round : 0 + const __m256i v_tmp_rnd = + _mm256_and_si256(_mm256_adds_epi16(v_abs_coeff, *v_round), v_zbin_mask); + // tmp32 = (int)(((((tmp * quant_ptr[rc != 0]) >> 16) + tmp) * + // quant_shift_ptr[rc != 0]) >> + // (16 - log_scale + AOM_QM_BITS)); + const __m256i v_tmp32_a = _mm256_mulhi_epi16(v_tmp_rnd, *v_quant); + const __m256i v_tmp32_b = _mm256_add_epi16(v_tmp32_a, v_tmp_rnd); + const __m256i v_tmp32_hi = _mm256_slli_epi16( + _mm256_mulhi_epi16(v_tmp32_b, *v_quant_shift), log_scale); + const __m256i v_tmp32_lo = _mm256_srli_epi16( + _mm256_mullo_epi16(v_tmp32_b, *v_quant_shift), 16 - log_scale); + const __m256i v_tmp32 = _mm256_or_si256(v_tmp32_hi, v_tmp32_lo); + const __m256i v_dqcoeff_hi = _mm256_slli_epi16( + _mm256_mulhi_epi16(v_tmp32, *v_dequant), 16 - log_scale); + const __m256i v_dqcoeff_lo = + _mm256_srli_epi16(_mm256_mullo_epi16(v_tmp32, *v_dequant), log_scale); + const __m256i v_dqcoeff = + _mm256_sign_epi16(_mm256_or_si256(v_dqcoeff_hi, v_dqcoeff_lo), v_coeff); + const __m256i v_qcoeff = _mm256_sign_epi16(v_tmp32, v_coeff); + const __m256i v_nz_mask = _mm256_cmpgt_epi16(v_tmp32, _mm256_setzero_si256()); + store_coefficients_avx2(v_qcoeff, qcoeff_ptr); + store_coefficients_avx2(v_dqcoeff, dqcoeff_ptr); + return v_nz_mask; +} + +static AOM_FORCE_INLINE void quantize_b_no_qmatrix_avx2( + const tran_low_t *coeff_ptr, intptr_t n_coeffs, const int16_t *zbin_ptr, + const int16_t *round_ptr, const int16_t *quant_ptr, + const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, + tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, + const int16_t *iscan, int log_scale) { + __m256i v_zbin, v_round, v_quant, v_dequant, v_quant_shift; + __m256i v_eobmax = _mm256_setzero_si256(); + + load_b_values_avx2(zbin_ptr, &v_zbin, round_ptr, &v_round, quant_ptr, + &v_quant, dequant_ptr, &v_dequant, quant_shift_ptr, + &v_quant_shift, log_scale); + + // Do DC and first 15 AC. + __m256i v_nz_mask = quantize_b_logscale_16( + coeff_ptr, qcoeff_ptr, dqcoeff_ptr, &v_quant, &v_dequant, &v_round, + &v_zbin, &v_quant_shift, log_scale); + + v_eobmax = get_max_lane_eob(iscan, v_eobmax, v_nz_mask); + + v_round = _mm256_unpackhi_epi64(v_round, v_round); + v_quant = _mm256_unpackhi_epi64(v_quant, v_quant); + v_dequant = _mm256_unpackhi_epi64(v_dequant, v_dequant); + v_quant_shift = _mm256_unpackhi_epi64(v_quant_shift, v_quant_shift); + v_zbin = _mm256_unpackhi_epi64(v_zbin, v_zbin); + + for (intptr_t count = n_coeffs - 16; count > 0; count -= 16) { + coeff_ptr += 16; + qcoeff_ptr += 16; + dqcoeff_ptr += 16; + iscan += 16; + v_nz_mask = quantize_b_logscale_16(coeff_ptr, qcoeff_ptr, dqcoeff_ptr, + &v_quant, &v_dequant, &v_round, &v_zbin, + &v_quant_shift, log_scale); + + v_eobmax = get_max_lane_eob(iscan, v_eobmax, v_nz_mask); + } + + *eob_ptr = accumulate_eob256(v_eobmax); +} + +void aom_quantize_b_32x32_avx2(const tran_low_t *coeff_ptr, intptr_t n_coeffs, + const int16_t *zbin_ptr, + const int16_t *round_ptr, + const int16_t *quant_ptr, + const int16_t *quant_shift_ptr, + tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, + const int16_t *dequant_ptr, uint16_t *eob_ptr, + const int16_t *scan, const int16_t *iscan) { + (void)scan; + quantize_b_no_qmatrix_avx2(coeff_ptr, n_coeffs, zbin_ptr, round_ptr, + quant_ptr, quant_shift_ptr, qcoeff_ptr, + dqcoeff_ptr, dequant_ptr, eob_ptr, iscan, 1); +} + +void aom_quantize_b_64x64_avx2(const tran_low_t *coeff_ptr, intptr_t n_coeffs, + const int16_t *zbin_ptr, + const int16_t *round_ptr, + const int16_t *quant_ptr, + const int16_t *quant_shift_ptr, + tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, + const int16_t *dequant_ptr, uint16_t *eob_ptr, + const int16_t *scan, const int16_t *iscan) { + (void)scan; + quantize_b_no_qmatrix_avx2(coeff_ptr, n_coeffs, zbin_ptr, round_ptr, + quant_ptr, quant_shift_ptr, qcoeff_ptr, + dqcoeff_ptr, dequant_ptr, eob_ptr, iscan, 2); +} |