Mercurial > libavcodec.hg
comparison x86/vp8dsp-init.c @ 12197:fbf4d5b1b664 libavcodec
Remove FF_MM_SSE2/3 flags for CPUs where this is generally not faster than
regular MMX code. Examples of this are the Core1 CPU. Instead, set a new flag,
FF_MM_SSE2/3SLOW, which can be checked for particular SSE2/3 functions that
have been checked specifically on such CPUs and are actually faster than
their MMX counterparts.
In addition, use this flag to enable particular VP8 and LPC SSE2 functions
that are faster than their MMX counterparts.
Based on a patch by Loren Merritt <lorenm AT u washington edu>.
| author | rbultje |
|---|---|
| date | Mon, 19 Jul 2010 22:38:23 +0000 |
| parents | 552c7c10bc73 |
| children | 677570e65a75 |
comparison
equal
deleted
inserted
replaced
| 12196:552c7c10bc73 | 12197:fbf4d5b1b664 |
|---|---|
| 326 if (mm_flags & FF_MM_SSE) { | 326 if (mm_flags & FF_MM_SSE) { |
| 327 c->put_vp8_epel_pixels_tab[0][0][0] = | 327 c->put_vp8_epel_pixels_tab[0][0][0] = |
| 328 c->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_sse; | 328 c->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_sse; |
| 329 } | 329 } |
| 330 | 330 |
| 331 if (mm_flags & FF_MM_SSE2) { | 331 if (mm_flags & (FF_MM_SSE2|FF_MM_SSE2SLOW)) { |
| 332 VP8_LUMA_MC_FUNC(0, 16, sse2); | 332 VP8_LUMA_MC_FUNC(0, 16, sse2); |
| 333 VP8_MC_FUNC(1, 8, sse2); | 333 VP8_MC_FUNC(1, 8, sse2); |
| 334 VP8_BILINEAR_MC_FUNC(0, 16, sse2); | 334 VP8_BILINEAR_MC_FUNC(0, 16, sse2); |
| 335 VP8_BILINEAR_MC_FUNC(1, 8, sse2); | 335 VP8_BILINEAR_MC_FUNC(1, 8, sse2); |
| 336 | 336 |
| 337 c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_sse2; | 337 c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_sse2; |
| 338 c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_sse2; | 338 c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_sse2; |
| 339 | 339 |
| 340 c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_sse2; | 340 c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_sse2; |
| 341 c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_sse2; | |
| 342 } | |
| 343 | |
| 344 if (mm_flags & FF_MM_SSE2) { | |
| 341 c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_sse2; | 345 c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_sse2; |
| 342 c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_sse2; | |
| 343 c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_sse2; | 346 c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_sse2; |
| 344 } | 347 } |
| 345 | 348 |
| 346 if (mm_flags & FF_MM_SSSE3) { | 349 if (mm_flags & FF_MM_SSSE3) { |
| 347 VP8_LUMA_MC_FUNC(0, 16, ssse3); | 350 VP8_LUMA_MC_FUNC(0, 16, ssse3); |