diff x86/vp8dsp-init.c @ 12340:2d15f62f4f8a libavcodec

VP8: move zeroing of luma DC block into the WHT Lets us do the zeroing in asm instead of C. Also makes it consistent with the way the regular iDCT code does it.
author darkshikari
date Mon, 02 Aug 2010 20:18:09 +0000
parents 435319d67bd8
children 3fc4c625b6f3
line wrap: on
line diff
--- a/x86/vp8dsp-init.c	Mon Aug 02 09:44:53 2010 +0000
+++ b/x86/vp8dsp-init.c	Mon Aug 02 20:18:09 2010 +0000
@@ -224,6 +224,7 @@
 extern void ff_vp8_idct_dc_add4y_sse2(uint8_t *dst, DCTELEM block[4][16], int stride);
 extern void ff_vp8_idct_dc_add4uv_mmx(uint8_t *dst, DCTELEM block[2][16], int stride);
 extern void ff_vp8_luma_dc_wht_mmx(DCTELEM block[4][4][16], DCTELEM dc[16]);
+extern void ff_vp8_luma_dc_wht_sse(DCTELEM block[4][4][16], DCTELEM dc[16]);
 extern void ff_vp8_idct_add_mmx(uint8_t *dst, DCTELEM block[16], int stride);
 extern void ff_vp8_idct_add_sse(uint8_t *dst, DCTELEM block[16], int stride);
 
@@ -335,6 +336,7 @@
 
     if (mm_flags & FF_MM_SSE) {
         c->vp8_idct_add                         = ff_vp8_idct_add_sse;
+        c->vp8_luma_dc_wht                      = ff_vp8_luma_dc_wht_sse;
         c->put_vp8_epel_pixels_tab[0][0][0]     =
         c->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_sse;
     }