| 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 <assert.h> | 11 #include <assert.h> |
| 12 #include <stdio.h> | 12 #include <stdio.h> |
| 13 | 13 |
| 14 #include "./vpx_dsp_rtcd.h" | 14 #include "./vpx_dsp_rtcd.h" |
| 15 #include "vpx_dsp/mips/vpx_common_dspr2.h" | 15 #include "vpx_dsp/mips/convolve_common_dspr2.h" |
| 16 #include "vpx_dsp/vpx_dsp_common.h" | 16 #include "vpx_dsp/vpx_dsp_common.h" |
| 17 #include "vpx_dsp/vpx_filter.h" | 17 #include "vpx_dsp/vpx_filter.h" |
| 18 #include "vpx_ports/mem.h" | 18 #include "vpx_ports/mem.h" |
| 19 | 19 |
| 20 #if HAVE_DSPR2 | 20 #if HAVE_DSPR2 |
| 21 uint8_t vpx_ff_cropTbl_a[256 + 2 * CROP_WIDTH]; | |
| 22 uint8_t *vpx_ff_cropTbl; | |
| 23 | |
| 24 void vpx_dsputil_static_init(void) { | |
| 25 int i; | |
| 26 | |
| 27 for (i = 0; i < 256; i++) vpx_ff_cropTbl_a[i + CROP_WIDTH] = i; | |
| 28 | |
| 29 for (i = 0; i < CROP_WIDTH; i++) { | |
| 30 vpx_ff_cropTbl_a[i] = 0; | |
| 31 vpx_ff_cropTbl_a[i + CROP_WIDTH + 256] = 255; | |
| 32 } | |
| 33 | |
| 34 vpx_ff_cropTbl = &vpx_ff_cropTbl_a[CROP_WIDTH]; | |
| 35 } | |
| 36 | |
| 37 static void convolve_horiz_4_transposed_dspr2(const uint8_t *src, | 21 static void convolve_horiz_4_transposed_dspr2(const uint8_t *src, |
| 38 int32_t src_stride, | 22 int32_t src_stride, |
| 39 uint8_t *dst, | 23 uint8_t *dst, |
| 40 int32_t dst_stride, | 24 int32_t dst_stride, |
| 41 const int16_t *filter_x0, | 25 const int16_t *filter_x0, |
| 42 int32_t h) { | 26 int32_t h) { |
| 43 int32_t y; | 27 int32_t y; |
| 44 uint8_t *cm = vpx_ff_cropTbl; | 28 uint8_t *cm = vpx_ff_cropTbl; |
| 45 uint8_t *dst_ptr; | 29 uint8_t *dst_ptr; |
| 46 int32_t vector1b, vector2b, vector3b, vector4b; | 30 int32_t vector1b, vector2b, vector3b, vector4b; |
| (...skipping 898 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 945 | 929 |
| 946 void vpx_convolve8_dspr2(const uint8_t *src, ptrdiff_t src_stride, | 930 void vpx_convolve8_dspr2(const uint8_t *src, ptrdiff_t src_stride, |
| 947 uint8_t *dst, ptrdiff_t dst_stride, | 931 uint8_t *dst, ptrdiff_t dst_stride, |
| 948 const int16_t *filter_x, int x_step_q4, | 932 const int16_t *filter_x, int x_step_q4, |
| 949 const int16_t *filter_y, int y_step_q4, | 933 const int16_t *filter_y, int y_step_q4, |
| 950 int w, int h) { | 934 int w, int h) { |
| 951 DECLARE_ALIGNED(32, uint8_t, temp[64 * 135]); | 935 DECLARE_ALIGNED(32, uint8_t, temp[64 * 135]); |
| 952 int32_t intermediate_height = ((h * y_step_q4) >> 4) + 7; | 936 int32_t intermediate_height = ((h * y_step_q4) >> 4) + 7; |
| 953 uint32_t pos = 38; | 937 uint32_t pos = 38; |
| 954 | 938 |
| 939 assert(x_step_q4 == 16); |
| 940 assert(y_step_q4 == 16); |
| 941 assert(((const int32_t *)filter_x)[1] != 0x800000); |
| 942 assert(((const int32_t *)filter_y)[1] != 0x800000); |
| 943 |
| 944 |
| 955 /* bit positon for extract from acc */ | 945 /* bit positon for extract from acc */ |
| 956 __asm__ __volatile__ ( | 946 __asm__ __volatile__ ( |
| 957 "wrdsp %[pos], 1 \n\t" | 947 "wrdsp %[pos], 1 \n\t" |
| 958 : | 948 : |
| 959 : [pos] "r" (pos) | 949 : [pos] "r" (pos) |
| 960 ); | 950 ); |
| 961 | 951 |
| 962 if (intermediate_height < h) | 952 if (intermediate_height < h) |
| 963 intermediate_height = h; | 953 intermediate_height = h; |
| 964 | 954 |
| 965 if (x_step_q4 != 16 || y_step_q4 != 16) | |
| 966 return vpx_convolve8_c(src, src_stride, | |
| 967 dst, dst_stride, | |
| 968 filter_x, x_step_q4, | |
| 969 filter_y, y_step_q4, | |
| 970 w, h); | |
| 971 | |
| 972 if ((((const int32_t *)filter_x)[1] == 0x800000) | |
| 973 && (((const int32_t *)filter_y)[1] == 0x800000)) | |
| 974 return vpx_convolve_copy(src, src_stride, | |
| 975 dst, dst_stride, | |
| 976 filter_x, x_step_q4, | |
| 977 filter_y, y_step_q4, | |
| 978 w, h); | |
| 979 | |
| 980 /* copy the src to dst */ | 955 /* copy the src to dst */ |
| 981 if (filter_x[3] == 0x80) { | 956 if (filter_x[3] == 0x80) { |
| 982 copy_horiz_transposed(src - src_stride * 3, src_stride, | 957 copy_horiz_transposed(src - src_stride * 3, src_stride, |
| 983 temp, intermediate_height, | 958 temp, intermediate_height, |
| 984 w, intermediate_height); | 959 w, intermediate_height); |
| 985 } else if (((const int32_t *)filter_x)[0] == 0) { | 960 } else if (((const int32_t *)filter_x)[0] == 0) { |
| 986 vpx_convolve2_dspr2(src - src_stride * 3, src_stride, | 961 vpx_convolve2_dspr2(src - src_stride * 3, src_stride, |
| 987 temp, intermediate_height, | 962 temp, intermediate_height, |
| 988 filter_x, | 963 filter_x, |
| 989 w, intermediate_height); | 964 w, intermediate_height); |
| (...skipping 283 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 1273 dst[x] = src[x]; | 1248 dst[x] = src[x]; |
| 1274 } | 1249 } |
| 1275 | 1250 |
| 1276 src += src_stride; | 1251 src += src_stride; |
| 1277 dst += dst_stride; | 1252 dst += dst_stride; |
| 1278 } | 1253 } |
| 1279 break; | 1254 break; |
| 1280 } | 1255 } |
| 1281 } | 1256 } |
| 1282 #endif | 1257 #endif |
| OLD | NEW |