OLD | NEW |
1 /* | 1 /* |
2 * Copyright (c) 2012 The WebM project authors. All Rights Reserved. | 2 * Copyright (c) 2012 The WebM project authors. All Rights Reserved. |
3 * | 3 * |
4 * Use of this source code is governed by a BSD-style license | 4 * Use of this source code is governed by a BSD-style license |
5 * that can be found in the LICENSE file in the root of the source | 5 * that can be found in the LICENSE file in the root of the source |
6 * tree. An additional intellectual property rights grant can be found | 6 * tree. An additional intellectual property rights grant can be found |
7 * in the file PATENTS. All contributing project authors may | 7 * in the file PATENTS. All contributing project authors may |
8 * be found in the AUTHORS file in the root of the source tree. | 8 * be found in the AUTHORS file in the root of the source tree. |
9 */ | 9 */ |
10 | 10 |
11 #include <immintrin.h> // AVX2 | 11 #include <immintrin.h> // AVX2 |
| 12 |
| 13 #include "./vp9_rtcd.h" |
12 #include "vp9/common/vp9_idct.h" // for cospi constants | 14 #include "vp9/common/vp9_idct.h" // for cospi constants |
13 #include "vpx_ports/mem.h" | 15 #include "vpx_ports/mem.h" |
14 | 16 |
15 #define pair256_set_epi16(a, b) \ | 17 #define pair256_set_epi16(a, b) \ |
16 _mm256_set_epi16((int16_t)(b), (int16_t)(a), (int16_t)(b), (int16_t)(a), \ | 18 _mm256_set_epi16((int16_t)(b), (int16_t)(a), (int16_t)(b), (int16_t)(a), \ |
17 (int16_t)(b), (int16_t)(a), (int16_t)(b), (int16_t)(a), \ | 19 (int16_t)(b), (int16_t)(a), (int16_t)(b), (int16_t)(a), \ |
18 (int16_t)(b), (int16_t)(a), (int16_t)(b), (int16_t)(a), \ | 20 (int16_t)(b), (int16_t)(a), (int16_t)(b), (int16_t)(a), \ |
19 (int16_t)(b), (int16_t)(a), (int16_t)(b), (int16_t)(a)) | 21 (int16_t)(b), (int16_t)(a), (int16_t)(b), (int16_t)(a)) |
20 | 22 |
21 #define pair256_set_epi32(a, b) \ | 23 #define pair256_set_epi32(a, b) \ |
(...skipping 2680 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
2702 _mm_storeu_si128((__m128i *)(output_nextStep + 6 * 32), _mm256_extract
f128_si256(tr2_6,1)); | 2704 _mm_storeu_si128((__m128i *)(output_nextStep + 6 * 32), _mm256_extract
f128_si256(tr2_6,1)); |
2703 _mm_storeu_si128((__m128i *)(output_nextStep + 7 * 32), _mm256_extract
f128_si256(tr2_7,1)); | 2705 _mm_storeu_si128((__m128i *)(output_nextStep + 7 * 32), _mm256_extract
f128_si256(tr2_7,1)); |
2704 // Process next 8x8 | 2706 // Process next 8x8 |
2705 output_currStep += 8; | 2707 output_currStep += 8; |
2706 output_nextStep += 8; | 2708 output_nextStep += 8; |
2707 } | 2709 } |
2708 } | 2710 } |
2709 } | 2711 } |
2710 } | 2712 } |
2711 } // NOLINT | 2713 } // NOLINT |
OLD | NEW |