Mercurial > libavcodec.hg
comparison dsputil.h @ 6384:0a403ade8c81 libavcodec
simd and unroll png_filter_row
cycles per 1000 pixels on core2:
left: 9211->5170
top: 9283->2138
avg: 12215->7611
paeth: 64024->17360
overall rgb png decoding speed: +45%
overall greyscale png decoding speed: +6%
author | lorenm |
---|---|
date | Thu, 21 Feb 2008 07:10:46 +0000 |
parents | 7ebd1cdb2142 |
children | 3dc36ec2dcad |
comparison
equal
deleted
inserted
replaced
6383:7ba06222bda7 | 6384:0a403ade8c81 |
---|---|
302 | 302 |
303 me_cmp_func pix_abs[2][4]; | 303 me_cmp_func pix_abs[2][4]; |
304 | 304 |
305 /* huffyuv specific */ | 305 /* huffyuv specific */ |
306 void (*add_bytes)(uint8_t *dst/*align 16*/, uint8_t *src/*align 16*/, int w); | 306 void (*add_bytes)(uint8_t *dst/*align 16*/, uint8_t *src/*align 16*/, int w); |
307 void (*add_bytes_l2)(uint8_t *dst/*align 16*/, uint8_t *src1/*align 16*/, uint8_t *src2/*align 16*/, int w); | |
307 void (*diff_bytes)(uint8_t *dst/*align 16*/, uint8_t *src1/*align 16*/, uint8_t *src2/*align 1*/,int w); | 308 void (*diff_bytes)(uint8_t *dst/*align 16*/, uint8_t *src1/*align 16*/, uint8_t *src2/*align 1*/,int w); |
308 /** | 309 /** |
309 * subtract huffyuv's variant of median prediction | 310 * subtract huffyuv's variant of median prediction |
310 * note, this might read from src1[-1], src2[-1] | 311 * note, this might read from src1[-1], src2[-1] |
311 */ | 312 */ |
312 void (*sub_hfyu_median_prediction)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w, int *left, int *left_top); | 313 void (*sub_hfyu_median_prediction)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w, int *left, int *left_top); |
314 /* this might write to dst[w] */ | |
315 void (*add_png_paeth_prediction)(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp); | |
313 void (*bswap_buf)(uint32_t *dst, const uint32_t *src, int w); | 316 void (*bswap_buf)(uint32_t *dst, const uint32_t *src, int w); |
314 | 317 |
315 void (*h264_v_loop_filter_luma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0); | 318 void (*h264_v_loop_filter_luma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0); |
316 void (*h264_h_loop_filter_luma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0); | 319 void (*h264_h_loop_filter_luma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0); |
317 void (*h264_v_loop_filter_chroma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0); | 320 void (*h264_v_loop_filter_chroma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0); |