diff vp8.c @ 12194:80b142c2e9f7 libavcodec

Change function prototypes for width=8 inner and mbedge loopfilter functions so that it does both U and V planes at the same time. This will have speed advantages when using SSE2 (or higher) optimizations, since we can do both the U and V rows together in a single xmm register. This also renames filter16 to filter16y and filter8 to filter8uv so that it's more obvious what each function is used for.

--- a/vp8.c	Mon Jul 19 20:53:58 2010 +0000
+++ b/vp8.c	Mon Jul 19 21:18:04 2010 +0000
@@ -1245,31 +1245,45 @@
      bedge_lim = 2* filter_level    + inner_limit;
 
     if (mb_x) {
-        s->vp8dsp.vp8_h_loop_filter16(dst[0], s->linesize,   mbedge_lim, inner_limit, hev_thresh);
-        s->vp8dsp.vp8_h_loop_filter8 (dst[1], s->uvlinesize, mbedge_lim, inner_limit, hev_thresh);
-        s->vp8dsp.vp8_h_loop_filter8 (dst[2], s->uvlinesize, mbedge_lim, inner_limit, hev_thresh);
+        s->vp8dsp.vp8_h_loop_filter16y(dst[0],     s->linesize,
+                                       mbedge_lim, inner_limit, hev_thresh);
+        s->vp8dsp.vp8_h_loop_filter8uv(dst[1],     dst[2],      s->uvlinesize,
+                                       mbedge_lim, inner_limit, hev_thresh);
     }
 
     if (!mb->skip || mb->mode == MODE_I4x4 || mb->mode == VP8_MVMODE_SPLIT) {
-        s->vp8dsp.vp8_h_loop_filter16_inner(dst[0]+ 4, s->linesize,   bedge_lim, inner_limit, hev_thresh);
-        s->vp8dsp.vp8_h_loop_filter16_inner(dst[0]+ 8, s->linesize,   bedge_lim, inner_limit, hev_thresh);
-        s->vp8dsp.vp8_h_loop_filter16_inner(dst[0]+12, s->linesize,   bedge_lim, inner_limit, hev_thresh);
-        s->vp8dsp.vp8_h_loop_filter8_inner (dst[1]+ 4, s->uvlinesize, bedge_lim, inner_limit, hev_thresh);
-        s->vp8dsp.vp8_h_loop_filter8_inner (dst[2]+ 4, s->uvlinesize, bedge_lim, inner_limit, hev_thresh);
+        s->vp8dsp.vp8_h_loop_filter16y_inner(dst[0]+ 4, s->linesize, bedge_lim,
+                                             inner_limit,   hev_thresh);
+        s->vp8dsp.vp8_h_loop_filter16y_inner(dst[0]+ 8, s->linesize, bedge_lim,
+                                             inner_limit,   hev_thresh);
+        s->vp8dsp.vp8_h_loop_filter16y_inner(dst[0]+12, s->linesize, bedge_lim,
+                                             inner_limit,   hev_thresh);
+        s->vp8dsp.vp8_h_loop_filter8uv_inner(dst[1] + 4,    dst[2] + 4,
+                                             s->uvlinesize, bedge_lim,
+                                             inner_limit,   hev_thresh);
     }
 
     if (mb_y) {
-        s->vp8dsp.vp8_v_loop_filter16(dst[0], s->linesize,   mbedge_lim, inner_limit, hev_thresh);
-        s->vp8dsp.vp8_v_loop_filter8 (dst[1], s->uvlinesize, mbedge_lim, inner_limit, hev_thresh);
-        s->vp8dsp.vp8_v_loop_filter8 (dst[2], s->uvlinesize, mbedge_lim, inner_limit, hev_thresh);
+        s->vp8dsp.vp8_v_loop_filter16y(dst[0],     s->linesize,
+                                       mbedge_lim, inner_limit, hev_thresh);
+        s->vp8dsp.vp8_v_loop_filter8uv(dst[1],     dst[2],      s->uvlinesize,
+                                       mbedge_lim, inner_limit, hev_thresh);
     }
 
     if (!mb->skip || mb->mode == MODE_I4x4 || mb->mode == VP8_MVMODE_SPLIT) {
-        s->vp8dsp.vp8_v_loop_filter16_inner(dst[0]+ 4*s->linesize,   s->linesize,   bedge_lim, inner_limit, hev_thresh);
-        s->vp8dsp.vp8_v_loop_filter16_inner(dst[0]+ 8*s->linesize,   s->linesize,   bedge_lim, inner_limit, hev_thresh);
-        s->vp8dsp.vp8_v_loop_filter16_inner(dst[0]+12*s->linesize,   s->linesize,   bedge_lim, inner_limit, hev_thresh);
-        s->vp8dsp.vp8_v_loop_filter8_inner (dst[1]+ 4*s->uvlinesize, s->uvlinesize, bedge_lim, inner_limit, hev_thresh);
-        s->vp8dsp.vp8_v_loop_filter8_inner (dst[2]+ 4*s->uvlinesize, s->uvlinesize, bedge_lim, inner_limit, hev_thresh);
+        s->vp8dsp.vp8_v_loop_filter16y_inner(dst[0]+ 4*s->linesize,
+                                             s->linesize,   bedge_lim,
+                                             inner_limit,   hev_thresh);
+        s->vp8dsp.vp8_v_loop_filter16y_inner(dst[0]+ 8*s->linesize,
+                                             s->linesize,   bedge_lim,
+                                             inner_limit,   hev_thresh);
+        s->vp8dsp.vp8_v_loop_filter16y_inner(dst[0]+12*s->linesize,
+                                             s->linesize,   bedge_lim,
+                                             inner_limit, hev_thresh);
+        s->vp8dsp.vp8_v_loop_filter8uv_inner(dst[1] + 4 * s->uvlinesize,
+                                             dst[2] + 4 * s->uvlinesize,
+                                             s->uvlinesize, bedge_lim,
+                                             inner_limit,   hev_thresh);
     }
 }