diff x86/vp8dsp-init.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/x86/vp8dsp-init.c	Mon Jul 19 20:53:58 2010 +0000
+++ b/x86/vp8dsp-init.c	Mon Jul 19 21:18:04 2010 +0000
@@ -230,18 +230,18 @@
 extern void ff_vp8_h_loop_filter_simple_mmxext(uint8_t *dst, int stride, int flim);
 extern void ff_vp8_h_loop_filter_simple_sse2  (uint8_t *dst, int stride, int flim);
 
-extern void ff_vp8_v_loop_filter16_inner_mmx   (uint8_t *dst, int stride,
-                                                int e, int i, int hvt);
-extern void ff_vp8_v_loop_filter16_inner_mmxext(uint8_t *dst, int stride,
-                                                int e, int i, int hvt);
-extern void ff_vp8_v_loop_filter16_inner_sse2  (uint8_t *dst, int stride,
-                                                int e, int i, int hvt);
-extern void ff_vp8_h_loop_filter16_inner_mmx   (uint8_t *dst, int stride,
-                                                int e, int i, int hvt);
-extern void ff_vp8_h_loop_filter16_inner_mmxext(uint8_t *dst, int stride,
-                                                int e, int i, int hvt);
-extern void ff_vp8_h_loop_filter16_inner_sse2  (uint8_t *dst, int stride,
-                                                int e, int i, int hvt);
+extern void ff_vp8_v_loop_filter16y_inner_mmx   (uint8_t *dst, int stride,
+                                                 int e, int i, int hvt);
+extern void ff_vp8_v_loop_filter16y_inner_mmxext(uint8_t *dst, int stride,
+                                                 int e, int i, int hvt);
+extern void ff_vp8_v_loop_filter16y_inner_sse2  (uint8_t *dst, int stride,
+                                                 int e, int i, int hvt);
+extern void ff_vp8_h_loop_filter16y_inner_mmx   (uint8_t *dst, int stride,
+                                                 int e, int i, int hvt);
+extern void ff_vp8_h_loop_filter16y_inner_mmxext(uint8_t *dst, int stride,
+                                                 int e, int i, int hvt);
+extern void ff_vp8_h_loop_filter16y_inner_sse2  (uint8_t *dst, int stride,
+                                                 int e, int i, int hvt);
 #endif
 
 #define VP8_LUMA_MC_FUNC(IDX, SIZE, OPT) \
@@ -284,8 +284,8 @@
         c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_mmx;
         c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_mmx;
 
-        c->vp8_v_loop_filter16_inner = ff_vp8_v_loop_filter16_inner_mmx;
-        c->vp8_h_loop_filter16_inner = ff_vp8_h_loop_filter16_inner_mmx;
+        c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_mmx;
+        c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_mmx;
     }
 
     /* note that 4-tap width=16 functions are missing because w=16
@@ -302,8 +302,8 @@
         c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_mmxext;
         c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_mmxext;
 
-        c->vp8_v_loop_filter16_inner = ff_vp8_v_loop_filter16_inner_mmxext;
-        c->vp8_h_loop_filter16_inner = ff_vp8_h_loop_filter16_inner_mmxext;
+        c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_mmxext;
+        c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_mmxext;
     }
 
     if (mm_flags & FF_MM_SSE) {
@@ -320,8 +320,8 @@
         c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_sse2;
         c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_sse2;
 
-        c->vp8_v_loop_filter16_inner = ff_vp8_v_loop_filter16_inner_sse2;
-        c->vp8_h_loop_filter16_inner = ff_vp8_h_loop_filter16_inner_sse2;
+        c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_sse2;
+        c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_sse2;
     }
 
     if (mm_flags & FF_MM_SSSE3) {