| OLD | NEW |
| 1 /* | 1 /* |
| 2 * Copyright (c) 2013 The WebM project authors. All Rights Reserved. | 2 * Copyright (c) 2013 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 "./vp9_rtcd.h" | 11 #include "./vp9_rtcd.h" |
| 12 #include "vp9/common/vp9_common.h" | 12 #include "vp9/common/vp9_common.h" |
| 13 #include "vpx_ports/mem.h" | 13 #include "vpx_ports/mem.h" |
| 14 | 14 |
| 15 void vp9_convolve8_neon(const uint8_t *src, ptrdiff_t src_stride, | 15 void vp9_convolve8_neon(const uint8_t *src, ptrdiff_t src_stride, |
| 16 uint8_t *dst, ptrdiff_t dst_stride, | 16 uint8_t *dst, ptrdiff_t dst_stride, |
| 17 const int16_t *filter_x, int x_step_q4, | 17 const int16_t *filter_x, int x_step_q4, |
| 18 const int16_t *filter_y, int y_step_q4, | 18 const int16_t *filter_y, int y_step_q4, |
| 19 int w, int h) { | 19 int w, int h) { |
| 20 /* Given our constraints: w <= 64, h <= 64, taps == 8 we can reduce the | 20 /* Given our constraints: w <= 64, h <= 64, taps == 8 we can reduce the |
| 21 * maximum buffer size to 64 * 64 + 7 (+ 1 to make it divisible by 4). | 21 * maximum buffer size to 64 * 64 + 7 (+ 1 to make it divisible by 4). |
| 22 */ | 22 */ |
| 23 DECLARE_ALIGNED_ARRAY(8, uint8_t, temp, 64 * 72); | 23 DECLARE_ALIGNED(8, uint8_t, temp[64 * 72]); |
| 24 | 24 |
| 25 // Account for the vertical phase needing 3 lines prior and 4 lines post | 25 // Account for the vertical phase needing 3 lines prior and 4 lines post |
| 26 int intermediate_height = h + 7; | 26 int intermediate_height = h + 7; |
| 27 | 27 |
| 28 if (x_step_q4 != 16 || y_step_q4 != 16) { | 28 if (x_step_q4 != 16 || y_step_q4 != 16) { |
| 29 vp9_convolve8_c(src, src_stride, | 29 vp9_convolve8_c(src, src_stride, |
| 30 dst, dst_stride, | 30 dst, dst_stride, |
| 31 filter_x, x_step_q4, | 31 filter_x, x_step_q4, |
| 32 filter_y, y_step_q4, | 32 filter_y, y_step_q4, |
| 33 w, h); | 33 w, h); |
| (...skipping 15 matching lines...) Expand all Loading... |
| 49 dst, dst_stride, | 49 dst, dst_stride, |
| 50 filter_x, x_step_q4, filter_y, y_step_q4, | 50 filter_x, x_step_q4, filter_y, y_step_q4, |
| 51 w, h); | 51 w, h); |
| 52 } | 52 } |
| 53 | 53 |
| 54 void vp9_convolve8_avg_neon(const uint8_t *src, ptrdiff_t src_stride, | 54 void vp9_convolve8_avg_neon(const uint8_t *src, ptrdiff_t src_stride, |
| 55 uint8_t *dst, ptrdiff_t dst_stride, | 55 uint8_t *dst, ptrdiff_t dst_stride, |
| 56 const int16_t *filter_x, int x_step_q4, | 56 const int16_t *filter_x, int x_step_q4, |
| 57 const int16_t *filter_y, int y_step_q4, | 57 const int16_t *filter_y, int y_step_q4, |
| 58 int w, int h) { | 58 int w, int h) { |
| 59 DECLARE_ALIGNED_ARRAY(8, uint8_t, temp, 64 * 72); | 59 DECLARE_ALIGNED(8, uint8_t, temp[64 * 72]); |
| 60 int intermediate_height = h + 7; | 60 int intermediate_height = h + 7; |
| 61 | 61 |
| 62 if (x_step_q4 != 16 || y_step_q4 != 16) { | 62 if (x_step_q4 != 16 || y_step_q4 != 16) { |
| 63 vp9_convolve8_avg_c(src, src_stride, | 63 vp9_convolve8_avg_c(src, src_stride, |
| 64 dst, dst_stride, | 64 dst, dst_stride, |
| 65 filter_x, x_step_q4, | 65 filter_x, x_step_q4, |
| 66 filter_y, y_step_q4, | 66 filter_y, y_step_q4, |
| 67 w, h); | 67 w, h); |
| 68 return; | 68 return; |
| 69 } | 69 } |
| 70 | 70 |
| 71 /* This implementation has the same issues as above. In addition, we only want | 71 /* This implementation has the same issues as above. In addition, we only want |
| 72 * to average the values after both passes. | 72 * to average the values after both passes. |
| 73 */ | 73 */ |
| 74 vp9_convolve8_horiz_neon(src - src_stride * 3, src_stride, | 74 vp9_convolve8_horiz_neon(src - src_stride * 3, src_stride, |
| 75 temp, 64, | 75 temp, 64, |
| 76 filter_x, x_step_q4, filter_y, y_step_q4, | 76 filter_x, x_step_q4, filter_y, y_step_q4, |
| 77 w, intermediate_height); | 77 w, intermediate_height); |
| 78 vp9_convolve8_avg_vert_neon(temp + 64 * 3, | 78 vp9_convolve8_avg_vert_neon(temp + 64 * 3, |
| 79 64, dst, dst_stride, | 79 64, dst, dst_stride, |
| 80 filter_x, x_step_q4, filter_y, y_step_q4, | 80 filter_x, x_step_q4, filter_y, y_step_q4, |
| 81 w, h); | 81 w, h); |
| 82 } | 82 } |
| OLD | NEW |