/* * Copyright (c) 2024, 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 #include #include "config/av1_rtcd.h" #include "av1/common/resize.h" #include "aom_dsp/x86/synonyms.h" #define ROW_OFFSET 5 #define CAST_HI(x) _mm256_castsi128_si256(x) #define CAST_LOW(x) _mm256_castsi256_si128(x) #define PROCESS_RESIZE_Y_WD16 \ const int idx1 = AOMMIN(height - 1, i + 5); \ const int idx2 = AOMMIN(height - 1, i + 6); \ l6 = l10; \ l7 = l11; \ l8 = _mm_loadu_si128((__m128i *)(data + idx1 * stride)); \ l9 = _mm_loadu_si128((__m128i *)(data + idx2 * stride)); \ \ /* g0... g15 | i0... i15 */ \ const __m256i s68 = \ _mm256_permute2x128_si256(CAST_HI(l6), CAST_HI(l8), 0x20); \ /* h0... h15 | j0... j15 */ \ const __m256i s79 = \ _mm256_permute2x128_si256(CAST_HI(l7), CAST_HI(l9), 0x20); \ \ /* g0h0... g7g7 | i0j0... i7j */ \ s[3] = _mm256_unpacklo_epi8(s68, s79); \ /* g8h8... g15g15 | i8j8... i15j15 */ \ s[8] = _mm256_unpackhi_epi8(s68, s79); \ \ __m256i res_out[2] = { 0 }; \ resize_convolve(s, coeffs_y, res_out); \ \ /* r00... r07 */ \ __m256i res_a_round_1 = _mm256_add_epi32(res_out[0], round_const_bits); \ /* r20... r27 */ \ __m256i res_a_round_2 = _mm256_add_epi32(res_out[1], round_const_bits); \ \ res_a_round_1 = _mm256_sra_epi32(res_a_round_1, round_shift_bits); \ res_a_round_2 = _mm256_sra_epi32(res_a_round_2, round_shift_bits); \ \ __m256i res_out_b[2] = { 0 }; \ resize_convolve(s + 5, coeffs_y, res_out_b); \ \ /* r08... r015 */ \ __m256i res_b_round_1 = _mm256_add_epi32(res_out_b[0], round_const_bits); \ /* r28... r215 */ \ __m256i res_b_round_2 = _mm256_add_epi32(res_out_b[1], round_const_bits); \ res_b_round_1 = _mm256_sra_epi32(res_b_round_1, round_shift_bits); \ res_b_round_2 = _mm256_sra_epi32(res_b_round_2, round_shift_bits); \ \ /* r00... r03 r20... r23 | r04... r07 r24... r27 */ \ __m256i res_8bit0 = _mm256_packus_epi32(res_a_round_1, res_a_round_2); \ /* r08... r012 r28... r212 | r013... r015 r213... r215 */ \ __m256i res_8bit1 = _mm256_packus_epi32(res_b_round_1, res_b_round_2); \ /* r00... r07 | r20... r27 */ \ res_8bit0 = _mm256_permute4x64_epi64(res_8bit0, 0xd8); \ /* r08... r015 | r28... r215 */ \ res_8bit1 = _mm256_permute4x64_epi64(res_8bit1, 0xd8); \ /* r00... r015 | r20... r215 */ \ res_8bit1 = _mm256_packus_epi16(res_8bit0, res_8bit1); \ res_8bit0 = _mm256_min_epu8(res_8bit1, clip_pixel); \ res_8bit0 = _mm256_max_epu8(res_8bit0, zero); #define PROCESS_RESIZE_Y_WD8 \ const int idx1 = AOMMIN(height - 1, i + 5); \ const int idx2 = AOMMIN(height - 1, i + 6); \ l6 = l10; \ l7 = l11; \ l8 = _mm_loadl_epi64((__m128i *)(data + idx1 * stride)); \ l9 = _mm_loadl_epi64((__m128i *)(data + idx2 * stride)); \ \ /* g0h0... g7h7 */ \ s67 = _mm_unpacklo_epi8(l6, l7); \ /* i0j0...i7j7 */ \ __m128i s89 = _mm_unpacklo_epi8(l8, l9); \ \ /* g0h0...g7g7 | i0j0...i7j7 */ \ s[3] = _mm256_permute2x128_si256(CAST_HI(s67), CAST_HI(s89), 0x20); \ \ __m256i res_out[2] = { 0 }; \ resize_convolve(s, coeffs_y, res_out); \ \ /* r00... r07 */ \ __m256i res_a_round_1 = _mm256_add_epi32(res_out[0], round_const_bits); \ /* r20...r27 */ \ __m256i res_a_round_2 = _mm256_add_epi32(res_out[1], round_const_bits); \ res_a_round_1 = _mm256_sra_epi32(res_a_round_1, round_shift_bits); \ res_a_round_2 = _mm256_sra_epi32(res_a_round_2, round_shift_bits); \ \ /* r00...r03 r20...r23 | r04...r07 r24...r27 */ \ res_a_round_1 = _mm256_packus_epi32(res_a_round_1, res_a_round_2); \ /* r00...r07 | r20...r27 */ \ res_a_round_1 = _mm256_permute4x64_epi64(res_a_round_1, 0xd8); \ res_a_round_1 = _mm256_packus_epi16(res_a_round_1, res_a_round_1); \ res_a_round_1 = _mm256_min_epu8(res_a_round_1, clip_pixel); \ res_a_round_1 = _mm256_max_epu8(res_a_round_1, zero); #define PROCESS_RESIZE_X_WD32 \ /* a0 a1 ..... a30 a31 */ \ __m256i row0 = _mm256_loadu_si256( \ (__m256i *)&input[i * in_stride + j - filter_offset]); \ /* b0 b1 ..... b30 b31 */ \ __m256i row1 = _mm256_loadu_si256( \ (__m256i *)&input[(i + 1) * in_stride + j - filter_offset]); \ /* a0 .... a15 || b0.... b15 */ \ __m256i r0 = _mm256_permute2x128_si256(row0, row1, 0x20); \ /* a16 .... a31 || b16 .... b31 */ \ __m256i r1 = _mm256_permute2x128_si256(row0, row1, 0x31); \ filter_offset = 3; \ \ /* Pad start pixels to the left, while processing the first pixels in the \ * row. */ \ if (j == 0) { \ /* a0 a0 a0 a0 .... a12 || b0 b0 b0 b0 .... b12 */ \ row0 = _mm256_shuffle_epi8(r0, wd32_start_pad_mask); \ /* a13 a14 a15 a16.....a28 || b13 b14 b15 b16.....b28 */ \ row1 = _mm256_alignr_epi8(r1, r0, 13); \ r0 = row0; \ r1 = row1; \ } \ const int is_last_cols32 = (j + 32 == filtered_length); \ /* Avoid loading extra pixels at frame boundary.*/ \ if (is_last_cols32) row_offset = ROW_OFFSET; \ /* a29 a30 a31 a32 a33 a34 a35 a36 0 0 ....*/ \ __m128i row0_0 = _mm_loadl_epi64( \ (__m128i *)&input[i * in_stride + 32 + j - filter_offset - row_offset]); \ /* b29 b30 b31 b32 b33 b34 b35 b36 0 0 .... */ \ __m128i row1_0 = \ _mm_loadl_epi64((__m128i *)&input[(i + 1) * in_stride + 32 + j - \ filter_offset - row_offset]); \ __m256i r2 = _mm256_permute2x128_si256( \ _mm256_castsi128_si256(row0_0), _mm256_castsi128_si256(row1_0), 0x20); \ \ /* Pad end pixels to the right, while processing the last pixels in the \ * row. */ \ if (is_last_cols32) { \ r2 = _mm256_shuffle_epi8(_mm256_srli_si256(r2, ROW_OFFSET), \ wd32_end_pad_mask); \ } \ \ /* Process even pixels of the first row */ \ /* a0 a0 a0 a0 a1 a2 .... a12 | b0 b0 b0 b0 b1 b2 .... b12 */ \ s0[0] = _mm256_alignr_epi8(r1, r0, 0); \ /* a0 a0 a1 a2 a3 a4 .... a14 | b0 b0 b1 b2 b3 b4 .... b14 */ \ s0[1] = _mm256_alignr_epi8(r1, r0, 2); \ /* a1 a2 a3 a4 a5 a6 .... a16 | b1 b2 b3 b4 b5 b6 .... b16 */ \ s0[2] = _mm256_alignr_epi8(r1, r0, 4); \ /* a3 a4 a5 a6 a7 a8 .... a18 | b3 b4 b5 b6 b7 b8 .... b18 */ \ s0[3] = _mm256_alignr_epi8(r1, r0, 6); \ \ /* Process even pixels of the second row */ \ /* a13 a14 a15 a16 ..... a28 | b13 b14 b15 b16 ..... b28 */ \ s1[0] = _mm256_alignr_epi8(r2, r1, 0); \ /* a15 a16 a17 a18 ..... a30 | b15 b16 b17 b18 ..... b30 */ \ s1[1] = _mm256_alignr_epi8(r2, r1, 2); \ /* a17 a18 a19 a20 ..... a32 | b17 b18 b19 b20 ..... b32 */ \ s1[2] = _mm256_alignr_epi8(r2, r1, 4); \ /* a19 a20 a21 a22 ..... a34 | b19 b20 b21 b22 ..... b34 */ \ s1[3] = _mm256_alignr_epi8(r2, r1, 6); \ \ /* The register res_out_0 stores the result of start-16 pixels corresponding \ * to the first and second rows whereas res_out_1 stores the end-16 \ * pixels. */ \ __m256i res_out_0[2], res_out_1[2]; \ res_out_1[0] = res_out_1[1] = zero; \ res_out_0[0] = res_out_0[1] = zero; \ resize_convolve(s0, coeffs_x, res_out_0); \ resize_convolve(s1, coeffs_x, res_out_1); \ \ /* Result of 32 pixels of row0 (a0 to a32) */ \ res_out_0[0] = _mm256_sra_epi32( \ _mm256_add_epi32(res_out_0[0], round_const_bits), round_shift_bits); \ res_out_1[0] = _mm256_sra_epi32( \ _mm256_add_epi32(res_out_1[0], round_const_bits), round_shift_bits); \ /* r00-r03 r08-r011 | r04-r07 r012-r015 */ \ __m256i res_out_r0 = _mm256_packus_epi32(res_out_0[0], res_out_1[0]); \ \ /* Result of 32 pixels of row1 (b0 to b32) */ \ res_out_0[1] = _mm256_sra_epi32( \ _mm256_add_epi32(res_out_0[1], round_const_bits), round_shift_bits); \ res_out_1[1] = _mm256_sra_epi32( \ _mm256_add_epi32(res_out_1[1], round_const_bits), round_shift_bits); \ /* r10-r13 r18-r111 | r14-r17 r112-r115 */ \ __m256i res_out_r1 = _mm256_packus_epi32(res_out_0[1], res_out_1[1]); \ \ /* Convert the result from 16bit to 8bit */ \ /* r00-r03 r08-r011 r10-r13 r18-r111 | r04-r07 r012-r015 r14-r17 r112-r115 \ */ \ __m256i res_out_r01 = _mm256_packus_epi16(res_out_r0, res_out_r1); \ __m256i res_out_row01 = _mm256_min_epu8(res_out_r01, clip_pixel); \ res_out_row01 = _mm256_max_epu8(res_out_r01, zero); \ __m128i low_128 = CAST_LOW(res_out_row01); \ __m128i high_128 = _mm256_extracti128_si256(res_out_row01, 1); \ \ _mm_storeu_si128((__m128i *)&intbuf[i * dst_stride + j / 2], \ _mm_unpacklo_epi32(low_128, high_128)); \ _mm_storeu_si128((__m128i *)&intbuf[(i + 1) * dst_stride + j / 2], \ _mm_unpackhi_epi32(low_128, high_128)); static inline void resize_convolve(const __m256i *const s, const __m256i *const coeffs, __m256i *res_out) { const __m256i res_0 = _mm256_maddubs_epi16(s[0], coeffs[0]); const __m256i res_1 = _mm256_maddubs_epi16(s[1], coeffs[1]); const __m256i res_2 = _mm256_maddubs_epi16(s[2], coeffs[2]); const __m256i res_3 = _mm256_maddubs_epi16(s[3], coeffs[3]); const __m256i dst_0 = _mm256_add_epi16(res_0, res_1); const __m256i dst_1 = _mm256_add_epi16(res_2, res_3); // The sum of convolve operation crosses signed 16bit. Hence, the addition // should happen in 32bit. const __m256i dst_00 = _mm256_cvtepi16_epi32(CAST_LOW(dst_0)); const __m256i dst_01 = _mm256_cvtepi16_epi32(_mm256_extracti128_si256(dst_0, 1)); const __m256i dst_10 = _mm256_cvtepi16_epi32(CAST_LOW(dst_1)); const __m256i dst_11 = _mm256_cvtepi16_epi32(_mm256_extracti128_si256(dst_1, 1)); res_out[0] = _mm256_add_epi32(dst_00, dst_10); res_out[1] = _mm256_add_epi32(dst_01, dst_11); } static inline void prepare_filter_coeffs(const int16_t *filter, __m256i *const coeffs /* [4] */) { // f0 f1 f2 f3 x x x x const __m128i sym_even_filter = _mm_loadl_epi64((__m128i *)filter); // f0 f1 f2 f3 f0 f1 f2 f3 const __m128i tmp0 = _mm_shuffle_epi32(sym_even_filter, 0x44); // f0 f1 f2 f3 f1 f0 f3 f2 const __m128i tmp1 = _mm_shufflehi_epi16(tmp0, 0xb1); const __m128i filter_8bit = _mm_packs_epi16(tmp1, tmp1); // f0 f1 f0 f1 .. coeffs[2] = _mm256_broadcastw_epi16(filter_8bit); // f2 f3 f2 f3 .. coeffs[3] = _mm256_broadcastw_epi16(_mm_bsrli_si128(filter_8bit, 2)); // f3 f2 f3 f2 .. coeffs[0] = _mm256_broadcastw_epi16(_mm_bsrli_si128(filter_8bit, 6)); // f1 f0 f1 f0 .. coeffs[1] = _mm256_broadcastw_epi16(_mm_bsrli_si128(filter_8bit, 4)); } bool av1_resize_vert_dir_avx2(uint8_t *intbuf, uint8_t *output, int out_stride, int height, int height2, int stride, int start_col) { assert(start_col <= stride); // For the GM tool, the input layer height or width is assured to be an even // number. Hence the function 'down2_symodd()' is not invoked and SIMD // optimization of the same is not implemented. // When the input height is less than 8 and even, the potential input // heights are limited to 2, 4, or 6. These scenarios require seperate // handling due to padding requirements. Invoking the C function here will // eliminate the need for conditional statements within the subsequent SIMD // code to manage these cases. if (height & 1 || height < 8) { return av1_resize_vert_dir_c(intbuf, output, out_stride, height, height2, stride, start_col); } __m256i s[10], coeffs_y[4]; const int bits = FILTER_BITS; const __m128i round_shift_bits = _mm_cvtsi32_si128(bits); const __m256i round_const_bits = _mm256_set1_epi32((1 << bits) >> 1); const uint8_t max_pixel = 255; const __m256i clip_pixel = _mm256_set1_epi8((char)max_pixel); const __m256i zero = _mm256_setzero_si256(); prepare_filter_coeffs(av1_down2_symeven_half_filter, coeffs_y); const int num_col16 = stride / 16; int remain_col = stride % 16; // The core vertical SIMD processes 4 input rows simultaneously to generate // output corresponding to 2 rows. To streamline the core loop and eliminate // the need for conditional checks, the remaining rows (4 or 6) are processed // separately. const int remain_row = (height % 4 == 0) ? 4 : 6; for (int j = start_col; j < stride - remain_col; j += 16) { const uint8_t *data = &intbuf[j]; const __m128i l3 = _mm_loadu_si128((__m128i *)(data + 0 * stride)); // Padding top 3 rows with the last available row at the top. const __m128i l0 = l3; const __m128i l1 = l3; const __m128i l2 = l3; const __m128i l4 = _mm_loadu_si128((__m128i *)(data + 1 * stride)); __m128i l6, l7, l8, l9; __m128i l5 = _mm_loadu_si128((__m128i *)(data + 2 * stride)); __m128i l10 = _mm_loadu_si128((__m128i *)(data + 3 * stride)); __m128i l11 = _mm_loadu_si128((__m128i *)(data + 4 * stride)); // a0...a15 | c0...c15 const __m256i s02 = _mm256_permute2x128_si256(CAST_HI(l0), CAST_HI(l2), 0x20); // b0...b15 | d0...d15 const __m256i s13 = _mm256_permute2x128_si256(CAST_HI(l1), CAST_HI(l3), 0x20); // c0...c15 | e0...e15 const __m256i s24 = _mm256_permute2x128_si256(CAST_HI(l2), CAST_HI(l4), 0x20); // d0...d15 | f0...f15 const __m256i s35 = _mm256_permute2x128_si256(CAST_HI(l3), CAST_HI(l5), 0x20); // e0...e15 | g0...g15 const __m256i s46 = _mm256_permute2x128_si256(CAST_HI(l4), CAST_HI(l10), 0x20); // f0...f15 | h0...h15 const __m256i s57 = _mm256_permute2x128_si256(CAST_HI(l5), CAST_HI(l11), 0x20); // a0b0...a7b7 | c0d0...c7d7 s[0] = _mm256_unpacklo_epi8(s02, s13); // c0d0...c7d7 | e0f0...e7f7 s[1] = _mm256_unpacklo_epi8(s24, s35); // e0f0...e7f7 | g0h0...g7h7 s[2] = _mm256_unpacklo_epi8(s46, s57); // a8b8...a15b15 | c8d8...c15d15 s[5] = _mm256_unpackhi_epi8(s02, s13); // c8d8...c15d15 | e8f8...e15f15 s[6] = _mm256_unpackhi_epi8(s24, s35); // e8f8...e15f15 | g8h8...g15h15 s[7] = _mm256_unpackhi_epi8(s46, s57); // height to be processed here const int process_ht = height - remain_row; for (int i = 0; i < process_ht; i += 4) { PROCESS_RESIZE_Y_WD16 _mm_storeu_si128((__m128i *)&output[(i / 2) * out_stride + j], CAST_LOW(res_8bit0)); _mm_storeu_si128( (__m128i *)&output[(i / 2) * out_stride + j + out_stride], _mm256_extracti128_si256(res_8bit0, 1)); // Load the required data for processing of next 4 input rows. const int idx7 = AOMMIN(height - 1, i + 7); const int idx8 = AOMMIN(height - 1, i + 8); l10 = _mm_loadu_si128((__m128i *)(data + idx7 * stride)); l11 = _mm_loadu_si128((__m128i *)(data + idx8 * stride)); const __m256i s810 = _mm256_permute2x128_si256(CAST_HI(l8), CAST_HI(l10), 0x20); const __m256i s911 = _mm256_permute2x128_si256(CAST_HI(l9), CAST_HI(l11), 0x20); // i0j0... i7j7 | k0l0... k7l7 s[4] = _mm256_unpacklo_epi8(s810, s911); // i8j8... i15j15 | k8l8... k15l15 s[9] = _mm256_unpackhi_epi8(s810, s911); s[0] = s[2]; s[1] = s[3]; s[2] = s[4]; s[5] = s[7]; s[6] = s[8]; s[7] = s[9]; } // Process the remaining last 4 or 6 rows here. int i = process_ht; while (i < height - 1) { PROCESS_RESIZE_Y_WD16 _mm_storeu_si128((__m128i *)&output[(i / 2) * out_stride + j], CAST_LOW(res_8bit0)); i += 2; const int is_store_valid = (i < height - 1); if (is_store_valid) _mm_storeu_si128((__m128i *)&output[(i / 2) * out_stride + j], _mm256_extracti128_si256(res_8bit0, 1)); i += 2; // Check if there is any remaining height to process. If so, perform the // necessary data loading for processing the next row. if (i < height - 1) { l10 = l11 = l9; const __m256i s810 = _mm256_permute2x128_si256(CAST_HI(l8), CAST_HI(l10), 0x20); const __m256i s911 = _mm256_permute2x128_si256(CAST_HI(l9), CAST_HI(l11), 0x20); // i0j0... i7j7 | k0l0... k7l7 s[4] = _mm256_unpacklo_epi8(s810, s911); // i8j8... i15j15 | k8l8... k15l15 s[9] = _mm256_unpackhi_epi8(s810, s911); s[0] = s[2]; s[1] = s[3]; s[2] = s[4]; s[5] = s[7]; s[6] = s[8]; s[7] = s[9]; } } } if (remain_col > 7) { const int processed_wd = num_col16 * 16; remain_col = stride % 8; const uint8_t *data = &intbuf[processed_wd]; const __m128i l3 = _mm_loadl_epi64((__m128i *)(data + 0 * stride)); // Padding top 3 rows with available top-most row. const __m128i l0 = l3; const __m128i l1 = l3; const __m128i l2 = l3; const __m128i l4 = _mm_loadl_epi64((__m128i *)(data + 1 * stride)); __m128i l6, l7, l8, l9; __m128i l5 = _mm_loadl_epi64((__m128i *)(data + 2 * stride)); __m128i l10 = _mm_loadl_epi64((__m128i *)(data + 3 * stride)); __m128i l11 = _mm_loadl_epi64((__m128i *)(data + 4 * stride)); // a0b0...a7b7 const __m128i s01 = _mm_unpacklo_epi8(l0, l1); // c0d0...c7d7 const __m128i s23 = _mm_unpacklo_epi8(l2, l3); // e0f0...e7f7 const __m128i s45 = _mm_unpacklo_epi8(l4, l5); // g0h0...g7h7 __m128i s67 = _mm_unpacklo_epi8(l10, l11); // a0b0...a7b7 | c0d0...c7d7 s[0] = _mm256_permute2x128_si256(CAST_HI(s01), CAST_HI(s23), 0x20); // c0d0...c7d7 | e0f0...e7f7 s[1] = _mm256_permute2x128_si256(CAST_HI(s23), CAST_HI(s45), 0x20); // e0f0...e7f7 | g0h0...g7h7 s[2] = _mm256_permute2x128_si256(CAST_HI(s45), CAST_HI(s67), 0x20); // height to be processed here const int process_ht = height - remain_row; for (int i = 0; i < process_ht; i += 4) { PROCESS_RESIZE_Y_WD8 _mm_storel_epi64((__m128i *)&output[(i / 2) * out_stride + processed_wd], CAST_LOW(res_a_round_1)); _mm_storel_epi64( (__m128i *)&output[(i / 2) * out_stride + processed_wd + out_stride], _mm256_extracti128_si256(res_a_round_1, 1)); const int idx7 = AOMMIN(height - 1, i + 7); const int idx8 = AOMMIN(height - 1, i + 8); l10 = _mm_loadl_epi64((__m128i *)(data + idx7 * stride)); l11 = _mm_loadl_epi64((__m128i *)(data + idx8 * stride)); // k0l0... k7l7 const __m128i s10s11 = _mm_unpacklo_epi8(l10, l11); // i0j0... i7j7 | k0l0... k7l7 s[4] = _mm256_permute2x128_si256(CAST_HI(s89), CAST_HI(s10s11), 0x20); s[0] = s[2]; s[1] = s[3]; s[2] = s[4]; } // Process the remaining last 4 or 6 rows here. int i = process_ht; while (i < height - 1) { PROCESS_RESIZE_Y_WD8 _mm_storel_epi64((__m128i *)&output[(i / 2) * out_stride + processed_wd], CAST_LOW(res_a_round_1)); i += 2; const int is_store_valid = (i < height - 1); if (is_store_valid) _mm_storel_epi64( (__m128i *)&output[(i / 2) * out_stride + processed_wd], _mm256_extracti128_si256(res_a_round_1, 1)); i += 2; // Check rows are still remaining for processing. If yes do the required // load of data for the next iteration. if (i < height - 1) { l10 = l11 = l9; // k0l0... k7l7 const __m128i s10s11 = _mm_unpacklo_epi8(l10, l11); // i0j0... i7j7 | k0l0... k7l7 s[4] = _mm256_permute2x128_si256(CAST_HI(s89), CAST_HI(s10s11), 0x20); s[0] = s[2]; s[1] = s[3]; s[2] = s[4]; } } } if (remain_col) return av1_resize_vert_dir_c(intbuf, output, out_stride, height, height2, stride, stride - remain_col); return true; } // Masks used for width 32 and 8 pixels, with left and right padding // requirements static const uint8_t wd32_left_padding_mask[32] = { 0, 0, 0, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 0, 0, 0, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12 }; static const uint8_t wd32_right_padding_mask[32] = { 0, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 0, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2 }; static const uint8_t wd8_right_padding_mask[32] = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 10, 10, 10, 10, 10, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 10, 10, 10, 10, 10 }; void av1_resize_horz_dir_avx2(const uint8_t *const input, int in_stride, uint8_t *intbuf, int height, int filtered_length, int width2) { assert(height % 2 == 0); // Invoke SSE2 for width less than 32. if (filtered_length < 32) { av1_resize_horz_dir_sse2(input, in_stride, intbuf, height, filtered_length, width2); return; } const int filt_length = sizeof(av1_down2_symeven_half_filter); assert(filt_length % 2 == 0); (void)filt_length; __m256i s0[4], s1[4], coeffs_x[4]; const int bits = FILTER_BITS; const int dst_stride = width2; const __m128i round_shift_bits = _mm_cvtsi32_si128(bits); const __m256i round_const_bits = _mm256_set1_epi32((1 << bits) >> 1); const uint8_t max_pixel = 255; const __m256i clip_pixel = _mm256_set1_epi8((char)max_pixel); const __m256i zero = _mm256_setzero_si256(); const __m256i wd32_start_pad_mask = _mm256_loadu_si256((__m256i *)wd32_left_padding_mask); const __m256i wd32_end_pad_mask = _mm256_loadu_si256((__m256i *)wd32_right_padding_mask); const __m256i wd8_end_pad_mask = _mm256_loadu_si256((__m256i *)wd8_right_padding_mask); prepare_filter_coeffs(av1_down2_symeven_half_filter, coeffs_x); // The core horizontal SIMD processes 32 input pixels of 2 rows simultaneously // to generate output corresponding to 2 rows. To streamline the core loop and // eliminate the need for conditional checks, the remaining columns (16 or 8) // are processed separately. if (filtered_length % 32 == 0) { for (int i = 0; i < height; i += 2) { int filter_offset = 0; int row_offset = 0; for (int j = 0; j < filtered_length; j += 32) { PROCESS_RESIZE_X_WD32 } } } else { for (int i = 0; i < height; i += 2) { int filter_offset = 0; int remain_col = filtered_length; int row_offset = 0; // To avoid pixel over-read at frame boundary, processing of 32 pixels // is done using the core loop only if sufficient number of pixels // required for the load are present. The remaining pixels are processed // separately. for (int j = 0; j <= filtered_length - 32; j += 32) { if (remain_col == 34 || remain_col == 36) { break; } PROCESS_RESIZE_X_WD32 remain_col -= 32; } int wd_processed = filtered_length - remain_col; // To avoid pixel over-read at frame boundary, processing of 16 pixels // is done only if sufficient number of pixels required for the // load are present. The remaining pixels are processed separately. if (remain_col > 15 && remain_col != 18 && remain_col != 20) { remain_col = filtered_length - wd_processed - 16; const int in_idx = i * in_stride + wd_processed; const int out_idx = (i * dst_stride) + wd_processed / 2; // a0 a1 --- a15 __m128i row0 = _mm_loadu_si128((__m128i *)&input[in_idx - filter_offset]); // b0 b1 --- b15 __m128i row1 = _mm_loadu_si128( (__m128i *)&input[in_idx + in_stride - filter_offset]); // a0 a1 --- a15 || b0 b1 --- b15 __m256i r0 = _mm256_permute2x128_si256(CAST_HI(row0), CAST_HI(row1), 0x20); if (filter_offset == 0) { r0 = _mm256_shuffle_epi8(r0, wd32_start_pad_mask); } filter_offset = 3; const int is_last_cols16 = wd_processed + 16 == filtered_length; if (is_last_cols16) row_offset = ROW_OFFSET; // a16 a17 --- a23 row0 = _mm_loadl_epi64( (__m128i *)&input[in_idx + 16 - row_offset - filter_offset]); // b16 b17 --- b23 row1 = _mm_loadl_epi64((__m128i *)&input[in_idx + 16 + in_stride - row_offset - filter_offset]); // a16-a23 x x x x| b16-b23 x x x x __m256i r1 = _mm256_permute2x128_si256(CAST_HI(row0), CAST_HI(row1), 0x20); // Pad end pixels to the right, while processing the last pixels in the // row. if (is_last_cols16) { r1 = _mm256_shuffle_epi8(_mm256_srli_si256(r1, ROW_OFFSET), wd32_end_pad_mask); } // a0 a1 --- a15 || b0 b1 --- b15 s0[0] = r0; // a2 a3 --- a17 || b2 b3 --- b17 s0[1] = _mm256_alignr_epi8(r1, r0, 2); // a4 a5 --- a19 || b4 b5 --- b19 s0[2] = _mm256_alignr_epi8(r1, r0, 4); // a6 a7 --- a21 || b6 b7 --- b21 s0[3] = _mm256_alignr_epi8(r1, r0, 6); // result for 16 pixels (a0 to a15) of row0 and row1 __m256i res_out_0[2]; res_out_0[0] = res_out_0[1] = zero; resize_convolve(s0, coeffs_x, res_out_0); // r00-r07 res_out_0[0] = _mm256_sra_epi32( _mm256_add_epi32(res_out_0[0], round_const_bits), round_shift_bits); // r10-r17 res_out_0[1] = _mm256_sra_epi32( _mm256_add_epi32(res_out_0[1], round_const_bits), round_shift_bits); // r00-r03 r10-r13 r04-r07 r14-r17 __m256i res_out_row01 = _mm256_packus_epi32(res_out_0[0], res_out_0[1]); // r00-r03 r10-r13 r00-r03 r10-r13 | r04-r07 r14-r17 r04-r07 r14-r17 res_out_row01 = _mm256_packus_epi16(res_out_row01, res_out_row01); res_out_row01 = _mm256_min_epu8(res_out_row01, clip_pixel); res_out_row01 = _mm256_max_epu8(res_out_row01, zero); // r00-r03 r10-r13 r04-r07 r14-r17 __m128i low_result = CAST_LOW(_mm256_permute4x64_epi64(res_out_row01, 0xd8)); // r00-r03 r04-r07 r10-r13 r14-r17 low_result = _mm_shuffle_epi32(low_result, 0xd8); _mm_storel_epi64((__m128i *)&intbuf[out_idx], low_result); _mm_storel_epi64((__m128i *)&intbuf[out_idx + dst_stride], _mm_unpackhi_epi64(low_result, low_result)); } // To avoid pixel over-read at frame boundary, processing of 8 pixels // is done only if sufficient number of pixels required for the // load are present. The remaining pixels are processed by C function. wd_processed = filtered_length - remain_col; if (remain_col > 7 && remain_col != 10 && remain_col != 12) { remain_col = filtered_length - wd_processed - 8; const int in_idx = i * in_stride + wd_processed - filter_offset; const int out_idx = (i * dst_stride) + wd_processed / 2; const int is_last_cols_8 = wd_processed + 8 == filtered_length; if (is_last_cols_8) row_offset = ROW_OFFSET; // a0 a1 --- a15 __m128i row0 = _mm_loadu_si128((__m128i *)&input[in_idx - row_offset]); // b0 b1 --- b15 __m128i row1 = _mm_loadu_si128((__m128i *)&input[in_idx + in_stride - row_offset]); // a0 a1 --- a15 || b0 b1 --- b15 __m256i r0 = _mm256_permute2x128_si256(CAST_HI(row0), CAST_HI(row1), 0x20); // Pad end pixels to the right, while processing the last pixels in the // row. if (is_last_cols_8) r0 = _mm256_shuffle_epi8(_mm256_srli_si256(r0, ROW_OFFSET), wd8_end_pad_mask); // a0 a1 a2 a3 a4 a5 a6 a7 | b0 b1 b2 b3 b4 b5 b6 b7 s0[0] = r0; // a2 a3 a4 a5 a6 a7 a8 a9 | b2 b3 b4 b5 b6 b7 b8 b9 s0[1] = _mm256_bsrli_epi128(r0, 2); // a4 a5 a6 a7 a8 a9 a10 a10 | b4 b5 b6 b7 b8 b9 b10 b10 s0[2] = _mm256_bsrli_epi128(r0, 4); // a6 a7 a8 a9 a10 a10 a10 a10 | b6 b7 b8 b9 b10 b10 b10 b10 s0[3] = _mm256_bsrli_epi128(r0, 6); __m256i res_out_0[2]; res_out_0[0] = res_out_0[1] = zero; resize_convolve(s0, coeffs_x, res_out_0); // r00 - r03 | r10 - r13 __m256i res_out = _mm256_permute2x128_si256(res_out_0[0], res_out_0[1], 0x20); // r00 - r03 | r10 - r13 res_out = _mm256_sra_epi32(_mm256_add_epi32(res_out, round_const_bits), round_shift_bits); // r00-r03 r00-r03 r10-r13 r10-r13 __m256i res_out_row01 = _mm256_packus_epi32(res_out, res_out); // r00-r03 r00-r03 r00-r03 r00-r03 r10-r13 r10-r13 r10-r13 r10-r13 res_out_row01 = _mm256_packus_epi16(res_out_row01, res_out_row01); res_out_row01 = _mm256_min_epu8(res_out_row01, clip_pixel); res_out_row01 = _mm256_max_epu8(res_out_row01, zero); xx_storel_32(intbuf + out_idx, CAST_LOW(res_out_row01)); xx_storel_32(intbuf + out_idx + dst_stride, _mm256_extracti128_si256(res_out_row01, 1)); } wd_processed = filtered_length - remain_col; if (remain_col) { const int in_idx = (in_stride * i); const int out_idx = (wd_processed / 2) + width2 * i; down2_symeven(input + in_idx, filtered_length, intbuf + out_idx, wd_processed); down2_symeven(input + in_idx + in_stride, filtered_length, intbuf + out_idx + width2, wd_processed); } } } }