[FFmpeg-devel] [PATCH 2/2] dsputil: move VC1-specific stuff into VC1DSPContext.

Ronald S. Bultje rsbultje
Thu Feb 17 20:54:48 CET 2011


---
 libavcodec/dsputil.c             |   53 ------------
 libavcodec/dsputil.h             |   27 ------
 libavcodec/ppc/dsputil_altivec.h |    1 -
 libavcodec/ppc/dsputil_ppc.c     |    2 -
 libavcodec/ppc/vc1dsp_altivec.c  |    6 +-
 libavcodec/vc1.c                 |   16 ++--
 libavcodec/vc1.h                 |    2 +
 libavcodec/vc1dec.c              |  174 +++++++++++++++++++-------------------
 libavcodec/vc1dsp.c              |   61 +++++++++++++-
 libavcodec/vc1dsp.h              |   67 +++++++++++++++
 libavcodec/x86/dsputil_mmx.c     |   20 -----
 libavcodec/x86/dsputil_mmx.h     |    1 -
 libavcodec/x86/vc1dsp_mmx.c      |   63 +++++++++-----
 13 files changed, 272 insertions(+), 221 deletions(-)
 create mode 100644 libavcodec/vc1dsp.h

diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c
index c89e58f..f7151b8 100644
--- a/libavcodec/dsputil.c
+++ b/libavcodec/dsputil.c
@@ -1600,54 +1600,6 @@ H264_CHROMA_MC(avg_       , op_avg)
 #undef op_avg
 #undef op_put
 
-static void put_no_rnd_vc1_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){
-    const int A=(8-x)*(8-y);
-    const int B=(  x)*(8-y);
-    const int C=(8-x)*(  y);
-    const int D=(  x)*(  y);
-    int i;
-
-    assert(x<8 && y<8 && x>=0 && y>=0);
-
-    for(i=0; i<h; i++)
-    {
-        dst[0] = (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + 32 - 4) >> 6;
-        dst[1] = (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + 32 - 4) >> 6;
-        dst[2] = (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + 32 - 4) >> 6;
-        dst[3] = (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + 32 - 4) >> 6;
-        dst[4] = (A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5] + 32 - 4) >> 6;
-        dst[5] = (A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6] + 32 - 4) >> 6;
-        dst[6] = (A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7] + 32 - 4) >> 6;
-        dst[7] = (A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8] + 32 - 4) >> 6;
-        dst+= stride;
-        src+= stride;
-    }
-}
-
-static void avg_no_rnd_vc1_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){
-    const int A=(8-x)*(8-y);
-    const int B=(  x)*(8-y);
-    const int C=(8-x)*(  y);
-    const int D=(  x)*(  y);
-    int i;
-
-    assert(x<8 && y<8 && x>=0 && y>=0);
-
-    for(i=0; i<h; i++)
-    {
-        dst[0] = avg2(dst[0], ((A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + 32 - 4) >> 6));
-        dst[1] = avg2(dst[1], ((A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + 32 - 4) >> 6));
-        dst[2] = avg2(dst[2], ((A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + 32 - 4) >> 6));
-        dst[3] = avg2(dst[3], ((A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + 32 - 4) >> 6));
-        dst[4] = avg2(dst[4], ((A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5] + 32 - 4) >> 6));
-        dst[5] = avg2(dst[5], ((A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6] + 32 - 4) >> 6));
-        dst[6] = avg2(dst[6], ((A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7] + 32 - 4) >> 6));
-        dst[7] = avg2(dst[7], ((A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8] + 32 - 4) >> 6));
-        dst+= stride;
-        src+= stride;
-    }
-}
-
 #define QPEL_MC(r, OPNAME, RND, OP) \
 static void OPNAME ## mpeg4_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\
     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
@@ -4301,17 +4253,12 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->avg_h264_chroma_pixels_tab[0]= avg_h264_chroma_mc8_c;
     c->avg_h264_chroma_pixels_tab[1]= avg_h264_chroma_mc4_c;
     c->avg_h264_chroma_pixels_tab[2]= avg_h264_chroma_mc2_c;
-    c->put_no_rnd_vc1_chroma_pixels_tab[0]= put_no_rnd_vc1_chroma_mc8_c;
-    c->avg_no_rnd_vc1_chroma_pixels_tab[0]= avg_no_rnd_vc1_chroma_mc8_c;
 
     c->draw_edges = draw_edges_c;
 
 #if CONFIG_MLP_DECODER || CONFIG_TRUEHD_DECODER
     ff_mlp_init(c, avctx);
 #endif
-#if CONFIG_VC1_DECODER
-    ff_vc1dsp_init(c,avctx);
-#endif
 #if CONFIG_WMV2_DECODER || CONFIG_VC1_DECODER
     ff_intrax8dsp_init(c,avctx);
 #endif
diff --git a/libavcodec/dsputil.h b/libavcodec/dsputil.h
index c811186..f97b2b5 100644
--- a/libavcodec/dsputil.h
+++ b/libavcodec/dsputil.h
@@ -341,9 +341,6 @@ typedef struct DSPContext {
      */
     h264_chroma_mc_func put_h264_chroma_pixels_tab[3];
     h264_chroma_mc_func avg_h264_chroma_pixels_tab[3];
-    /* This is really one func used in VC-1 decoding */
-    h264_chroma_mc_func put_no_rnd_vc1_chroma_pixels_tab[3];
-    h264_chroma_mc_func avg_no_rnd_vc1_chroma_pixels_tab[3];
 
     qpel_mc_func put_h264_qpel_pixels_tab[4][16];
     qpel_mc_func avg_h264_qpel_pixels_tab[4][16];
@@ -503,29 +500,6 @@ typedef struct DSPContext {
                                unsigned int filter_shift, int32_t mask, int blocksize,
                                int32_t *sample_buffer);
 
-    /* vc1 functions */
-    void (*vc1_inv_trans_8x8)(DCTELEM *b);
-    void (*vc1_inv_trans_8x4)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_inv_trans_4x8)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_inv_trans_4x4)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_inv_trans_8x8_dc)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_inv_trans_8x4_dc)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_inv_trans_4x8_dc)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_inv_trans_4x4_dc)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_v_overlap)(uint8_t* src, int stride);
-    void (*vc1_h_overlap)(uint8_t* src, int stride);
-    void (*vc1_v_loop_filter4)(uint8_t *src, int stride, int pq);
-    void (*vc1_h_loop_filter4)(uint8_t *src, int stride, int pq);
-    void (*vc1_v_loop_filter8)(uint8_t *src, int stride, int pq);
-    void (*vc1_h_loop_filter8)(uint8_t *src, int stride, int pq);
-    void (*vc1_v_loop_filter16)(uint8_t *src, int stride, int pq);
-    void (*vc1_h_loop_filter16)(uint8_t *src, int stride, int pq);
-    /* put 8x8 block with bicubic interpolation and quarterpel precision
-     * last argument is actually round value instead of height
-     */
-    op_pixels_func put_vc1_mspel_pixels_tab[16];
-    op_pixels_func avg_vc1_mspel_pixels_tab[16];
-
     /* intrax8 functions */
     void (*x8_spatial_compensation[12])(uint8_t *src , uint8_t *dst, int linesize);
     void (*x8_setup_spatial_compensation)(uint8_t *src, uint8_t *dst, int linesize,
@@ -629,7 +603,6 @@ void dsputil_init_vis(DSPContext* c, AVCodecContext *avctx);
 void ff_dsputil_init_dwt(DSPContext *c);
 void ff_rv30dsp_init(DSPContext* c, AVCodecContext *avctx);
 void ff_rv40dsp_init(DSPContext* c, AVCodecContext *avctx);
-void ff_vc1dsp_init(DSPContext* c, AVCodecContext *avctx);
 void ff_intrax8dsp_init(DSPContext* c, AVCodecContext *avctx);
 void ff_mlp_init(DSPContext* c, AVCodecContext *avctx);
 void ff_mlp_init_x86(DSPContext* c, AVCodecContext *avctx);
diff --git a/libavcodec/ppc/dsputil_altivec.h b/libavcodec/ppc/dsputil_altivec.h
index cd44f60..4147eec 100644
--- a/libavcodec/ppc/dsputil_altivec.h
+++ b/libavcodec/ppc/dsputil_altivec.h
@@ -43,7 +43,6 @@ void ff_vp3_idct_add_altivec(uint8_t *dest, int line_size, DCTELEM *block);
 void dsputil_h264_init_ppc(DSPContext* c, AVCodecContext *avctx);
 
 void dsputil_init_altivec(DSPContext* c, AVCodecContext *avctx);
-void vc1dsp_init_altivec(DSPContext* c, AVCodecContext *avctx);
 void float_init_altivec(DSPContext* c, AVCodecContext *avctx);
 void int_init_altivec(DSPContext* c, AVCodecContext *avctx);
 
diff --git a/libavcodec/ppc/dsputil_ppc.c b/libavcodec/ppc/dsputil_ppc.c
index 9e4f1aa..d13c1ce 100644
--- a/libavcodec/ppc/dsputil_ppc.c
+++ b/libavcodec/ppc/dsputil_ppc.c
@@ -171,8 +171,6 @@ void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx)
 
     if (av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC) {
         dsputil_init_altivec(c, avctx);
-        if(CONFIG_VC1_DECODER)
-            vc1dsp_init_altivec(c, avctx);
         float_init_altivec(c, avctx);
         int_init_altivec(c, avctx);
         c->gmc1 = gmc1_altivec;
diff --git a/libavcodec/ppc/vc1dsp_altivec.c b/libavcodec/ppc/vc1dsp_altivec.c
index 5ffe9a5..9577438 100644
--- a/libavcodec/ppc/vc1dsp_altivec.c
+++ b/libavcodec/ppc/vc1dsp_altivec.c
@@ -322,7 +322,11 @@ static void vc1_inv_trans_8x4_altivec(uint8_t *dest, int stride, DCTELEM *block)
 }
 
 
-void vc1dsp_init_altivec(DSPContext* dsp, AVCodecContext *avctx) {
+void ff_vc1dsp_init_altivec(VC1DSPContext* dsp)
+{
+    if (!(av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC))
+        return;
+
     dsp->vc1_inv_trans_8x8 = vc1_inv_trans_8x8_altivec;
     dsp->vc1_inv_trans_8x4 = vc1_inv_trans_8x4_altivec;
 }
diff --git a/libavcodec/vc1.c b/libavcodec/vc1.c
index 1f39878..8bd6647 100644
--- a/libavcodec/vc1.c
+++ b/libavcodec/vc1.c
@@ -337,14 +337,14 @@ int vc1_decode_sequence_header(AVCodecContext *avctx, VC1Context *v, GetBitConte
     v->res_fasttx = get_bits1(gb);
     if (!v->res_fasttx)
     {
-        v->s.dsp.vc1_inv_trans_8x8 = ff_simple_idct;
-        v->s.dsp.vc1_inv_trans_8x4 = ff_simple_idct84_add;
-        v->s.dsp.vc1_inv_trans_4x8 = ff_simple_idct48_add;
-        v->s.dsp.vc1_inv_trans_4x4 = ff_simple_idct44_add;
-        v->s.dsp.vc1_inv_trans_8x8_dc = ff_simple_idct_add;
-        v->s.dsp.vc1_inv_trans_8x4_dc = ff_simple_idct84_add;
-        v->s.dsp.vc1_inv_trans_4x8_dc = ff_simple_idct48_add;
-        v->s.dsp.vc1_inv_trans_4x4_dc = ff_simple_idct44_add;
+        v->vc1dsp.vc1_inv_trans_8x8 = ff_simple_idct;
+        v->vc1dsp.vc1_inv_trans_8x4 = ff_simple_idct84_add;
+        v->vc1dsp.vc1_inv_trans_4x8 = ff_simple_idct48_add;
+        v->vc1dsp.vc1_inv_trans_4x4 = ff_simple_idct44_add;
+        v->vc1dsp.vc1_inv_trans_8x8_dc = ff_simple_idct_add;
+        v->vc1dsp.vc1_inv_trans_8x4_dc = ff_simple_idct84_add;
+        v->vc1dsp.vc1_inv_trans_4x8_dc = ff_simple_idct48_add;
+        v->vc1dsp.vc1_inv_trans_4x4_dc = ff_simple_idct44_add;
     }
 
     v->fastuvmc =  get_bits1(gb); //common
diff --git a/libavcodec/vc1.h b/libavcodec/vc1.h
index da0b6c1..d826c5a 100644
--- a/libavcodec/vc1.h
+++ b/libavcodec/vc1.h
@@ -26,6 +26,7 @@
 #include "avcodec.h"
 #include "mpegvideo.h"
 #include "intrax8.h"
+#include "vc1dsp.h"
 
 /** Markers used in VC-1 AP frame data */
 //@{
@@ -155,6 +156,7 @@ enum COTypes {
 typedef struct VC1Context{
     MpegEncContext s;
     IntraX8Context x8;
+    VC1DSPContext vc1dsp;
 
     int bits;
 
diff --git a/libavcodec/vc1dec.c b/libavcodec/vc1dec.c
index e254d1b..6707cde 100644
--- a/libavcodec/vc1dec.c
+++ b/libavcodec/vc1dec.c
@@ -160,29 +160,30 @@ enum Imode {
 
 /** @} */ //Bitplane group
 
-static void vc1_loop_filter_iblk(MpegEncContext *s, int pq)
+static void vc1_loop_filter_iblk(VC1Context *v, int pq)
 {
+    MpegEncContext *s = &v->s;
     int j;
     if (!s->first_slice_line) {
-        s->dsp.vc1_v_loop_filter16(s->dest[0], s->linesize, pq);
+        v->vc1dsp.vc1_v_loop_filter16(s->dest[0], s->linesize, pq);
         if (s->mb_x)
-            s->dsp.vc1_h_loop_filter16(s->dest[0] - 16*s->linesize, s->linesize, pq);
-        s->dsp.vc1_h_loop_filter16(s->dest[0] - 16*s->linesize+8, s->linesize, pq);
+            v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16*s->linesize, s->linesize, pq);
+        v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16*s->linesize+8, s->linesize, pq);
         for(j = 0; j < 2; j++){
-            s->dsp.vc1_v_loop_filter8(s->dest[j+1], s->uvlinesize, pq);
+            v->vc1dsp.vc1_v_loop_filter8(s->dest[j+1], s->uvlinesize, pq);
             if (s->mb_x)
-                s->dsp.vc1_h_loop_filter8(s->dest[j+1]-8*s->uvlinesize, s->uvlinesize, pq);
+                v->vc1dsp.vc1_h_loop_filter8(s->dest[j+1]-8*s->uvlinesize, s->uvlinesize, pq);
         }
     }
-    s->dsp.vc1_v_loop_filter16(s->dest[0] + 8*s->linesize, s->linesize, pq);
+    v->vc1dsp.vc1_v_loop_filter16(s->dest[0] + 8*s->linesize, s->linesize, pq);
 
     if (s->mb_y == s->mb_height-1) {
         if (s->mb_x) {
-            s->dsp.vc1_h_loop_filter16(s->dest[0], s->linesize, pq);
-            s->dsp.vc1_h_loop_filter8(s->dest[1], s->uvlinesize, pq);
-            s->dsp.vc1_h_loop_filter8(s->dest[2], s->uvlinesize, pq);
+            v->vc1dsp.vc1_h_loop_filter16(s->dest[0], s->linesize, pq);
+            v->vc1dsp.vc1_h_loop_filter8(s->dest[1], s->uvlinesize, pq);
+            v->vc1dsp.vc1_h_loop_filter8(s->dest[2], s->uvlinesize, pq);
         }
-        s->dsp.vc1_h_loop_filter16(s->dest[0] + 8, s->linesize, pq);
+        v->vc1dsp.vc1_h_loop_filter16(s->dest[0] + 8, s->linesize, pq);
     }
 }
 
@@ -342,11 +343,11 @@ static void vc1_mc_1mv(VC1Context *v, int dir)
 
     if(s->mspel) {
         dxy = ((my & 3) << 2) | (mx & 3);
-        dsp->put_vc1_mspel_pixels_tab[dxy](s->dest[0]    , srcY    , s->linesize, v->rnd);
-        dsp->put_vc1_mspel_pixels_tab[dxy](s->dest[0] + 8, srcY + 8, s->linesize, v->rnd);
+        v->vc1dsp.put_vc1_mspel_pixels_tab[dxy](s->dest[0]    , srcY    , s->linesize, v->rnd);
+        v->vc1dsp.put_vc1_mspel_pixels_tab[dxy](s->dest[0] + 8, srcY + 8, s->linesize, v->rnd);
         srcY += s->linesize * 8;
-        dsp->put_vc1_mspel_pixels_tab[dxy](s->dest[0] + 8 * s->linesize    , srcY    , s->linesize, v->rnd);
-        dsp->put_vc1_mspel_pixels_tab[dxy](s->dest[0] + 8 * s->linesize + 8, srcY + 8, s->linesize, v->rnd);
+        v->vc1dsp.put_vc1_mspel_pixels_tab[dxy](s->dest[0] + 8 * s->linesize    , srcY    , s->linesize, v->rnd);
+        v->vc1dsp.put_vc1_mspel_pixels_tab[dxy](s->dest[0] + 8 * s->linesize + 8, srcY + 8, s->linesize, v->rnd);
     } else { // hpel mc - always used for luma
         dxy = (my & 2) | ((mx & 2) >> 1);
 
@@ -364,8 +365,8 @@ static void vc1_mc_1mv(VC1Context *v, int dir)
         dsp->put_h264_chroma_pixels_tab[0](s->dest[1], srcU, s->uvlinesize, 8, uvmx, uvmy);
         dsp->put_h264_chroma_pixels_tab[0](s->dest[2], srcV, s->uvlinesize, 8, uvmx, uvmy);
     }else{
-        dsp->put_no_rnd_vc1_chroma_pixels_tab[0](s->dest[1], srcU, s->uvlinesize, 8, uvmx, uvmy);
-        dsp->put_no_rnd_vc1_chroma_pixels_tab[0](s->dest[2], srcV, s->uvlinesize, 8, uvmx, uvmy);
+        v->vc1dsp.put_no_rnd_vc1_chroma_pixels_tab[0](s->dest[1], srcU, s->uvlinesize, 8, uvmx, uvmy);
+        v->vc1dsp.put_no_rnd_vc1_chroma_pixels_tab[0](s->dest[2], srcV, s->uvlinesize, 8, uvmx, uvmy);
     }
 }
 
@@ -433,7 +434,7 @@ static void vc1_mc_4mv_luma(VC1Context *v, int n)
 
     if(s->mspel) {
         dxy = ((my & 3) << 2) | (mx & 3);
-        dsp->put_vc1_mspel_pixels_tab[dxy](s->dest[0] + off, srcY, s->linesize, v->rnd);
+        v->vc1dsp.put_vc1_mspel_pixels_tab[dxy](s->dest[0] + off, srcY, s->linesize, v->rnd);
     } else { // hpel mc - always used for luma
         dxy = (my & 2) | ((mx & 2) >> 1);
         if(!v->rnd)
@@ -583,8 +584,8 @@ static void vc1_mc_4mv_chroma(VC1Context *v)
         dsp->put_h264_chroma_pixels_tab[0](s->dest[1], srcU, s->uvlinesize, 8, uvmx, uvmy);
         dsp->put_h264_chroma_pixels_tab[0](s->dest[2], srcV, s->uvlinesize, 8, uvmx, uvmy);
     }else{
-        dsp->put_no_rnd_vc1_chroma_pixels_tab[0](s->dest[1], srcU, s->uvlinesize, 8, uvmx, uvmy);
-        dsp->put_no_rnd_vc1_chroma_pixels_tab[0](s->dest[2], srcV, s->uvlinesize, 8, uvmx, uvmy);
+        v->vc1dsp.put_no_rnd_vc1_chroma_pixels_tab[0](s->dest[1], srcU, s->uvlinesize, 8, uvmx, uvmy);
+        v->vc1dsp.put_no_rnd_vc1_chroma_pixels_tab[0](s->dest[2], srcV, s->uvlinesize, 8, uvmx, uvmy);
     }
 }
 
@@ -906,11 +907,11 @@ static void vc1_interp_mc(VC1Context *v)
 
     if(s->mspel) {
         dxy = ((my & 3) << 2) | (mx & 3);
-        dsp->avg_vc1_mspel_pixels_tab[dxy](s->dest[0]    , srcY    , s->linesize, v->rnd);
-        dsp->avg_vc1_mspel_pixels_tab[dxy](s->dest[0] + 8, srcY + 8, s->linesize, v->rnd);
+        v->vc1dsp.avg_vc1_mspel_pixels_tab[dxy](s->dest[0]    , srcY    , s->linesize, v->rnd);
+        v->vc1dsp.avg_vc1_mspel_pixels_tab[dxy](s->dest[0] + 8, srcY + 8, s->linesize, v->rnd);
         srcY += s->linesize * 8;
-        dsp->avg_vc1_mspel_pixels_tab[dxy](s->dest[0] + 8 * s->linesize    , srcY    , s->linesize, v->rnd);
-        dsp->avg_vc1_mspel_pixels_tab[dxy](s->dest[0] + 8 * s->linesize + 8, srcY + 8, s->linesize, v->rnd);
+        v->vc1dsp.avg_vc1_mspel_pixels_tab[dxy](s->dest[0] + 8 * s->linesize    , srcY    , s->linesize, v->rnd);
+        v->vc1dsp.avg_vc1_mspel_pixels_tab[dxy](s->dest[0] + 8 * s->linesize + 8, srcY + 8, s->linesize, v->rnd);
     } else { // hpel mc
         dxy = (my & 2) | ((mx & 2) >> 1);
 
@@ -928,8 +929,8 @@ static void vc1_interp_mc(VC1Context *v)
         dsp->avg_h264_chroma_pixels_tab[0](s->dest[1], srcU, s->uvlinesize, 8, uvmx, uvmy);
         dsp->avg_h264_chroma_pixels_tab[0](s->dest[2], srcV, s->uvlinesize, 8, uvmx, uvmy);
     }else{
-        dsp->avg_no_rnd_vc1_chroma_pixels_tab[0](s->dest[1], srcU, s->uvlinesize, 8, uvmx, uvmy);
-        dsp->avg_no_rnd_vc1_chroma_pixels_tab[0](s->dest[2], srcV, s->uvlinesize, 8, uvmx, uvmy);
+        v->vc1dsp.avg_no_rnd_vc1_chroma_pixels_tab[0](s->dest[1], srcU, s->uvlinesize, 8, uvmx, uvmy);
+        v->vc1dsp.avg_no_rnd_vc1_chroma_pixels_tab[0](s->dest[2], srcV, s->uvlinesize, 8, uvmx, uvmy);
     }
 }
 
@@ -2039,15 +2040,15 @@ static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquan
         }
         if(!skip_block){
             if(i==1)
-                s->dsp.vc1_inv_trans_8x8_dc(dst, linesize, block);
+                v->vc1dsp.vc1_inv_trans_8x8_dc(dst, linesize, block);
             else{
-                s->dsp.vc1_inv_trans_8x8(block);
+                v->vc1dsp.vc1_inv_trans_8x8(block);
                 s->dsp.add_pixels_clamped(block, dst, linesize);
             }
             if(apply_filter && cbp_top  & 0xC)
-                s->dsp.vc1_v_loop_filter8(dst, linesize, v->pq);
+                v->vc1dsp.vc1_v_loop_filter8(dst, linesize, v->pq);
             if(apply_filter && cbp_left & 0xA)
-                s->dsp.vc1_h_loop_filter8(dst, linesize, v->pq);
+                v->vc1dsp.vc1_h_loop_filter8(dst, linesize, v->pq);
         }
         break;
     case TT_4X4:
@@ -2068,13 +2069,13 @@ static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquan
             }
             if(!(subblkpat & (1 << (3 - j))) && !skip_block){
                 if(i==1)
-                    s->dsp.vc1_inv_trans_4x4_dc(dst + (j&1)*4 + (j&2)*2*linesize, linesize, block + off);
+                    v->vc1dsp.vc1_inv_trans_4x4_dc(dst + (j&1)*4 + (j&2)*2*linesize, linesize, block + off);
                 else
-                    s->dsp.vc1_inv_trans_4x4(dst + (j&1)*4 + (j&2)*2*linesize, linesize, block + off);
+                    v->vc1dsp.vc1_inv_trans_4x4(dst + (j&1)*4 + (j&2)*2*linesize, linesize, block + off);
                 if(apply_filter && (j&2 ? pat & (1<<(j-2)) : (cbp_top & (1 << (j + 2)))))
-                    s->dsp.vc1_v_loop_filter4(dst + (j&1)*4 + (j&2)*2*linesize, linesize, v->pq);
+                    v->vc1dsp.vc1_v_loop_filter4(dst + (j&1)*4 + (j&2)*2*linesize, linesize, v->pq);
                 if(apply_filter && (j&1 ? pat & (1<<(j-1)) : (cbp_left & (1 << (j + 1)))))
-                    s->dsp.vc1_h_loop_filter4(dst + (j&1)*4 + (j&2)*2*linesize, linesize, v->pq);
+                    v->vc1dsp.vc1_h_loop_filter4(dst + (j&1)*4 + (j&2)*2*linesize, linesize, v->pq);
             }
         }
         break;
@@ -2096,13 +2097,13 @@ static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquan
             }
             if(!(subblkpat & (1 << (1 - j))) && !skip_block){
                 if(i==1)
-                    s->dsp.vc1_inv_trans_8x4_dc(dst + j*4*linesize, linesize, block + off);
+                    v->vc1dsp.vc1_inv_trans_8x4_dc(dst + j*4*linesize, linesize, block + off);
                 else
-                    s->dsp.vc1_inv_trans_8x4(dst + j*4*linesize, linesize, block + off);
+                    v->vc1dsp.vc1_inv_trans_8x4(dst + j*4*linesize, linesize, block + off);
                 if(apply_filter && j ? pat & 0x3 : (cbp_top & 0xC))
-                    s->dsp.vc1_v_loop_filter8(dst + j*4*linesize, linesize, v->pq);
+                    v->vc1dsp.vc1_v_loop_filter8(dst + j*4*linesize, linesize, v->pq);
                 if(apply_filter && cbp_left & (2 << j))
-                    s->dsp.vc1_h_loop_filter4(dst + j*4*linesize, linesize, v->pq);
+                    v->vc1dsp.vc1_h_loop_filter4(dst + j*4*linesize, linesize, v->pq);
             }
         }
         break;
@@ -2124,13 +2125,13 @@ static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquan
             }
             if(!(subblkpat & (1 << (1 - j))) && !skip_block){
                 if(i==1)
-                    s->dsp.vc1_inv_trans_4x8_dc(dst + j*4, linesize, block + off);
+                    v->vc1dsp.vc1_inv_trans_4x8_dc(dst + j*4, linesize, block + off);
                 else
-                    s->dsp.vc1_inv_trans_4x8(dst + j*4, linesize, block + off);
+                    v->vc1dsp.vc1_inv_trans_4x8(dst + j*4, linesize, block + off);
                 if(apply_filter && cbp_top & (2 << j))
-                    s->dsp.vc1_v_loop_filter4(dst + j*4, linesize, v->pq);
+                    v->vc1dsp.vc1_v_loop_filter4(dst + j*4, linesize, v->pq);
                 if(apply_filter && j ? pat & 0x5 : (cbp_left & 0xA))
-                    s->dsp.vc1_h_loop_filter8(dst + j*4, linesize, v->pq);
+                    v->vc1dsp.vc1_h_loop_filter8(dst + j*4, linesize, v->pq);
             }
         }
         break;
@@ -2232,14 +2233,14 @@ static int vc1_decode_p_mb(VC1Context *v)
 
                     vc1_decode_intra_block(v, s->block[i], i, val, mquant, (i&4)?v->codingset2:v->codingset);
                     if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue;
-                    s->dsp.vc1_inv_trans_8x8(s->block[i]);
+                    v->vc1dsp.vc1_inv_trans_8x8(s->block[i]);
                     if(v->rangeredfrm) for(j = 0; j < 64; j++) s->block[i][j] <<= 1;
                     s->dsp.put_signed_pixels_clamped(s->block[i], s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
                     if(v->pq >= 9 && v->overlap) {
                         if(v->c_avail)
-                            s->dsp.vc1_h_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
+                            v->vc1dsp.vc1_h_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
                         if(v->a_avail)
-                            s->dsp.vc1_v_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
+                            v->vc1dsp.vc1_v_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
                     }
                     if(apply_loop_filter && s->mb_x && s->mb_x != (s->mb_width - 1) && s->mb_y && s->mb_y != (s->mb_height - 1)){
                         int left_cbp, top_cbp;
@@ -2251,9 +2252,9 @@ static int vc1_decode_p_mb(VC1Context *v)
                             top_cbp  = (i & 2) ? (cbp >> ((i-2)*4)) : (v->cbp[s->mb_x - s->mb_stride] >> ((i+2)*4));
                         }
                         if(left_cbp & 0xC)
-                            s->dsp.vc1_v_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
+                            v->vc1dsp.vc1_v_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
                         if(top_cbp  & 0xA)
-                            s->dsp.vc1_h_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
+                            v->vc1dsp.vc1_h_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
                     }
                     block_cbp |= 0xF << (i << 2);
                 } else if(val) {
@@ -2268,9 +2269,9 @@ static int vc1_decode_p_mb(VC1Context *v)
                             top_cbp  = (i & 2) ? (cbp >> ((i-2)*4)) : (v->cbp[s->mb_x - s->mb_stride] >> ((i+2)*4));
                         }
                         if(left_cbp & 0xC)
-                            s->dsp.vc1_v_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
+                            v->vc1dsp.vc1_v_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
                         if(top_cbp  & 0xA)
-                            s->dsp.vc1_h_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
+                            v->vc1dsp.vc1_h_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
                     }
                     pat = vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block, s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize, (i&4) && (s->flags & CODEC_FLAG_GRAY), filter, left_cbp, top_cbp);
                     block_cbp |= pat << (i << 2);
@@ -2363,14 +2364,14 @@ static int vc1_decode_p_mb(VC1Context *v)
 
                     vc1_decode_intra_block(v, s->block[i], i, is_coded[i], mquant, (i&4)?v->codingset2:v->codingset);
                     if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue;
-                    s->dsp.vc1_inv_trans_8x8(s->block[i]);
+                    v->vc1dsp.vc1_inv_trans_8x8(s->block[i]);
                     if(v->rangeredfrm) for(j = 0; j < 64; j++) s->block[i][j] <<= 1;
                     s->dsp.put_signed_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize);
                     if(v->pq >= 9 && v->overlap) {
                         if(v->c_avail)
-                            s->dsp.vc1_h_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
+                            v->vc1dsp.vc1_h_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
                         if(v->a_avail)
-                            s->dsp.vc1_v_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
+                            v->vc1dsp.vc1_v_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
                     }
                     if(v->s.loop_filter && s->mb_x && s->mb_x != (s->mb_width - 1) && s->mb_y && s->mb_y != (s->mb_height - 1)){
                         int left_cbp, top_cbp;
@@ -2382,9 +2383,9 @@ static int vc1_decode_p_mb(VC1Context *v)
                             top_cbp  = (i & 2) ? (cbp >> ((i-2)*4)) : (v->cbp[s->mb_x - s->mb_stride] >> ((i+2)*4));
                         }
                         if(left_cbp & 0xC)
-                            s->dsp.vc1_v_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
+                            v->vc1dsp.vc1_v_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
                         if(top_cbp  & 0xA)
-                            s->dsp.vc1_h_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
+                            v->vc1dsp.vc1_h_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
                     }
                     block_cbp |= 0xF << (i << 2);
                 } else if(is_coded[i]) {
@@ -2399,9 +2400,9 @@ static int vc1_decode_p_mb(VC1Context *v)
                             top_cbp  = (i & 2) ? (cbp >> ((i-2)*4)) : (v->cbp[s->mb_x - s->mb_stride] >> ((i+2)*4));
                         }
                         if(left_cbp & 0xC)
-                            s->dsp.vc1_v_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
+                            v->vc1dsp.vc1_v_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
                         if(top_cbp  & 0xA)
-                            s->dsp.vc1_h_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
+                            v->vc1dsp.vc1_h_loop_filter8(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize, v->pq);
                     }
                     pat = vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block, s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize, (i&4) && (s->flags & CODEC_FLAG_GRAY), filter, left_cbp, top_cbp);
                     block_cbp |= pat << (i << 2);
@@ -2568,7 +2569,7 @@ static void vc1_decode_b_mb(VC1Context *v)
 
             vc1_decode_intra_block(v, s->block[i], i, val, mquant, (i&4)?v->codingset2:v->codingset);
             if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue;
-            s->dsp.vc1_inv_trans_8x8(s->block[i]);
+            v->vc1dsp.vc1_inv_trans_8x8(s->block[i]);
             if(v->rangeredfrm) for(j = 0; j < 64; j++) s->block[i][j] <<= 1;
             s->dsp.put_signed_pixels_clamped(s->block[i], s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
         } else if(val) {
@@ -2650,7 +2651,7 @@ static void vc1_decode_i_blocks(VC1Context *v)
 
                 vc1_decode_i_block(v, s->block[k], k, val, (k<4)? v->codingset : v->codingset2);
 
-                s->dsp.vc1_inv_trans_8x8(s->block[k]);
+                v->vc1dsp.vc1_inv_trans_8x8(s->block[k]);
                 if(v->pq >= 9 && v->overlap) {
                     for(j = 0; j < 64; j++) s->block[k][j] += 128;
                 }
@@ -2659,27 +2660,27 @@ static void vc1_decode_i_blocks(VC1Context *v)
             vc1_put_block(v, s->block);
             if(v->pq >= 9 && v->overlap) {
                 if(s->mb_x) {
-                    s->dsp.vc1_h_overlap(s->dest[0], s->linesize);
-                    s->dsp.vc1_h_overlap(s->dest[0] + 8 * s->linesize, s->linesize);
+                    v->vc1dsp.vc1_h_overlap(s->dest[0], s->linesize);
+                    v->vc1dsp.vc1_h_overlap(s->dest[0] + 8 * s->linesize, s->linesize);
                     if(!(s->flags & CODEC_FLAG_GRAY)) {
-                        s->dsp.vc1_h_overlap(s->dest[1], s->uvlinesize);
-                        s->dsp.vc1_h_overlap(s->dest[2], s->uvlinesize);
+                        v->vc1dsp.vc1_h_overlap(s->dest[1], s->uvlinesize);
+                        v->vc1dsp.vc1_h_overlap(s->dest[2], s->uvlinesize);
                     }
                 }
-                s->dsp.vc1_h_overlap(s->dest[0] + 8, s->linesize);
-                s->dsp.vc1_h_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize);
+                v->vc1dsp.vc1_h_overlap(s->dest[0] + 8, s->linesize);
+                v->vc1dsp.vc1_h_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize);
                 if(!s->first_slice_line) {
-                    s->dsp.vc1_v_overlap(s->dest[0], s->linesize);
-                    s->dsp.vc1_v_overlap(s->dest[0] + 8, s->linesize);
+                    v->vc1dsp.vc1_v_overlap(s->dest[0], s->linesize);
+                    v->vc1dsp.vc1_v_overlap(s->dest[0] + 8, s->linesize);
                     if(!(s->flags & CODEC_FLAG_GRAY)) {
-                        s->dsp.vc1_v_overlap(s->dest[1], s->uvlinesize);
-                        s->dsp.vc1_v_overlap(s->dest[2], s->uvlinesize);
+                        v->vc1dsp.vc1_v_overlap(s->dest[1], s->uvlinesize);
+                        v->vc1dsp.vc1_v_overlap(s->dest[2], s->uvlinesize);
                     }
                 }
-                s->dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize, s->linesize);
-                s->dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize);
+                v->vc1dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize, s->linesize);
+                v->vc1dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize);
             }
-            if(v->s.loop_filter) vc1_loop_filter_iblk(s, v->pq);
+            if(v->s.loop_filter) vc1_loop_filter_iblk(v, v->pq);
 
             if(get_bits_count(&s->gb) > v->bits) {
                 ff_er_add_slice(s, 0, 0, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END));
@@ -2790,34 +2791,34 @@ static void vc1_decode_i_blocks_adv(VC1Context *v)
 
                 vc1_decode_i_block_adv(v, s->block[k], k, val, (k<4)? v->codingset : v->codingset2, mquant);
 
-                s->dsp.vc1_inv_trans_8x8(s->block[k]);
+                v->vc1dsp.vc1_inv_trans_8x8(s->block[k]);
                 for(j = 0; j < 64; j++) s->block[k][j] += 128;
             }
 
             vc1_put_block(v, s->block);
             if(overlap) {
                 if(s->mb_x) {
-                    s->dsp.vc1_h_overlap(s->dest[0], s->linesize);
-                    s->dsp.vc1_h_overlap(s->dest[0] + 8 * s->linesize, s->linesize);
+                    v->vc1dsp.vc1_h_overlap(s->dest[0], s->linesize);
+                    v->vc1dsp.vc1_h_overlap(s->dest[0] + 8 * s->linesize, s->linesize);
                     if(!(s->flags & CODEC_FLAG_GRAY)) {
-                        s->dsp.vc1_h_overlap(s->dest[1], s->uvlinesize);
-                        s->dsp.vc1_h_overlap(s->dest[2], s->uvlinesize);
+                        v->vc1dsp.vc1_h_overlap(s->dest[1], s->uvlinesize);
+                        v->vc1dsp.vc1_h_overlap(s->dest[2], s->uvlinesize);
                     }
                 }
-                s->dsp.vc1_h_overlap(s->dest[0] + 8, s->linesize);
-                s->dsp.vc1_h_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize);
+                v->vc1dsp.vc1_h_overlap(s->dest[0] + 8, s->linesize);
+                v->vc1dsp.vc1_h_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize);
                 if(!s->first_slice_line) {
-                    s->dsp.vc1_v_overlap(s->dest[0], s->linesize);
-                    s->dsp.vc1_v_overlap(s->dest[0] + 8, s->linesize);
+                    v->vc1dsp.vc1_v_overlap(s->dest[0], s->linesize);
+                    v->vc1dsp.vc1_v_overlap(s->dest[0] + 8, s->linesize);
                     if(!(s->flags & CODEC_FLAG_GRAY)) {
-                        s->dsp.vc1_v_overlap(s->dest[1], s->uvlinesize);
-                        s->dsp.vc1_v_overlap(s->dest[2], s->uvlinesize);
+                        v->vc1dsp.vc1_v_overlap(s->dest[1], s->uvlinesize);
+                        v->vc1dsp.vc1_v_overlap(s->dest[2], s->uvlinesize);
                     }
                 }
-                s->dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize, s->linesize);
-                s->dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize);
+                v->vc1dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize, s->linesize);
+                v->vc1dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize);
             }
-            if(v->s.loop_filter) vc1_loop_filter_iblk(s, v->pq);
+            if(v->s.loop_filter) vc1_loop_filter_iblk(v, v->pq);
 
             if(get_bits_count(&s->gb) > v->bits) {
                 ff_er_add_slice(s, 0, 0, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END));
@@ -2929,7 +2930,7 @@ static void vc1_decode_b_blocks(VC1Context *v)
                 av_log(s->avctx, AV_LOG_ERROR, "Bits overconsumption: %i > %i at %ix%i\n", get_bits_count(&s->gb), v->bits,s->mb_x,s->mb_y);
                 return;
             }
-            if(v->s.loop_filter) vc1_loop_filter_iblk(s, v->pq);
+            if(v->s.loop_filter) vc1_loop_filter_iblk(v, v->pq);
         }
         if (!v->s.loop_filter)
             ff_draw_horiz_band(s, s->mb_y * 16, 16);
@@ -3023,6 +3024,7 @@ static av_cold int vc1_decode_init(AVCodecContext *avctx)
     if(ff_msmpeg4_decode_init(avctx) < 0)
         return -1;
     if (vc1_init_common(v) < 0) return -1;
+    ff_vc1dsp_init(&v->vc1dsp);
     for (i = 0; i < 64;  i++) {
 #define transpose(x) ((x>>3) | ((x&7)<<3))
         v->zz_8x8[0][i] = transpose(wmv1_scantable[0][i]);
diff --git a/libavcodec/vc1dsp.c b/libavcodec/vc1dsp.c
index fd740e1..000dad7 100644
--- a/libavcodec/vc1dsp.c
+++ b/libavcodec/vc1dsp.c
@@ -25,7 +25,7 @@
  *
  */
 
-#include "dsputil.h"
+#include "vc1dsp.h"
 
 
 /** Apply overlap transform to horizontal edge
@@ -612,7 +612,56 @@ PUT_VC1_MSPEL(1, 3)
 PUT_VC1_MSPEL(2, 3)
 PUT_VC1_MSPEL(3, 3)
 
-av_cold void ff_vc1dsp_init(DSPContext* dsp, AVCodecContext *avctx) {
+static void put_no_rnd_vc1_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){
+    const int A=(8-x)*(8-y);
+    const int B=(  x)*(8-y);
+    const int C=(8-x)*(  y);
+    const int D=(  x)*(  y);
+    int i;
+
+    assert(x<8 && y<8 && x>=0 && y>=0);
+
+    for(i=0; i<h; i++)
+    {
+        dst[0] = (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + 32 - 4) >> 6;
+        dst[1] = (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + 32 - 4) >> 6;
+        dst[2] = (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + 32 - 4) >> 6;
+        dst[3] = (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + 32 - 4) >> 6;
+        dst[4] = (A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5] + 32 - 4) >> 6;
+        dst[5] = (A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6] + 32 - 4) >> 6;
+        dst[6] = (A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7] + 32 - 4) >> 6;
+        dst[7] = (A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8] + 32 - 4) >> 6;
+        dst+= stride;
+        src+= stride;
+    }
+}
+
+#define avg2(a,b) ((a+b+1)>>1)
+static void avg_no_rnd_vc1_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){
+    const int A=(8-x)*(8-y);
+    const int B=(  x)*(8-y);
+    const int C=(8-x)*(  y);
+    const int D=(  x)*(  y);
+    int i;
+
+    assert(x<8 && y<8 && x>=0 && y>=0);
+
+    for(i=0; i<h; i++)
+    {
+        dst[0] = avg2(dst[0], ((A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + 32 - 4) >> 6));
+        dst[1] = avg2(dst[1], ((A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + 32 - 4) >> 6));
+        dst[2] = avg2(dst[2], ((A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + 32 - 4) >> 6));
+        dst[3] = avg2(dst[3], ((A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + 32 - 4) >> 6));
+        dst[4] = avg2(dst[4], ((A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5] + 32 - 4) >> 6));
+        dst[5] = avg2(dst[5], ((A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6] + 32 - 4) >> 6));
+        dst[6] = avg2(dst[6], ((A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7] + 32 - 4) >> 6));
+        dst[7] = avg2(dst[7], ((A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8] + 32 - 4) >> 6));
+        dst+= stride;
+        src+= stride;
+    }
+}
+
+av_cold void ff_vc1dsp_init(VC1DSPContext* dsp) {
     dsp->vc1_inv_trans_8x8 = vc1_inv_trans_8x8_c;
     dsp->vc1_inv_trans_4x8 = vc1_inv_trans_4x8_c;
     dsp->vc1_inv_trans_8x4 = vc1_inv_trans_8x4_c;
@@ -663,4 +712,12 @@ av_cold void ff_vc1dsp_init(DSPContext* dsp, AVCodecContext *avctx) {
     dsp->avg_vc1_mspel_pixels_tab[13] = avg_vc1_mspel_mc13_c;
     dsp->avg_vc1_mspel_pixels_tab[14] = avg_vc1_mspel_mc23_c;
     dsp->avg_vc1_mspel_pixels_tab[15] = avg_vc1_mspel_mc33_c;
+
+    dsp->put_no_rnd_vc1_chroma_pixels_tab[0]= put_no_rnd_vc1_chroma_mc8_c;
+    dsp->avg_no_rnd_vc1_chroma_pixels_tab[0]= avg_no_rnd_vc1_chroma_mc8_c;
+
+    if (HAVE_ALTIVEC)
+        ff_vc1dsp_init_altivec(dsp);
+    if (HAVE_MMX)
+        ff_vc1dsp_init_mmx(dsp);
 }
diff --git a/libavcodec/vc1dsp.h b/libavcodec/vc1dsp.h
new file mode 100644
index 0000000..a1f3d90
--- /dev/null
+++ b/libavcodec/vc1dsp.h
@@ -0,0 +1,67 @@
+/*
+ * VC-1 and WMV3 decoder - DSP functions
+ * Copyright (c) 2006 Konstantin Shishkov
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/**
+ * @file
+ * VC-1 and WMV3 decoder
+ *
+ */
+
+#ifndef AVCODEC_VC1DSP_H
+#define AVCODEC_VC1DSP_H
+
+#include "dsputil.h"
+
+typedef struct VC1DSPContext {
+    /* vc1 functions */
+    void (*vc1_inv_trans_8x8)(DCTELEM *b);
+    void (*vc1_inv_trans_8x4)(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*vc1_inv_trans_4x8)(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*vc1_inv_trans_4x4)(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*vc1_inv_trans_8x8_dc)(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*vc1_inv_trans_8x4_dc)(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*vc1_inv_trans_4x8_dc)(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*vc1_inv_trans_4x4_dc)(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*vc1_v_overlap)(uint8_t* src, int stride);
+    void (*vc1_h_overlap)(uint8_t* src, int stride);
+    void (*vc1_v_loop_filter4)(uint8_t *src, int stride, int pq);
+    void (*vc1_h_loop_filter4)(uint8_t *src, int stride, int pq);
+    void (*vc1_v_loop_filter8)(uint8_t *src, int stride, int pq);
+    void (*vc1_h_loop_filter8)(uint8_t *src, int stride, int pq);
+    void (*vc1_v_loop_filter16)(uint8_t *src, int stride, int pq);
+    void (*vc1_h_loop_filter16)(uint8_t *src, int stride, int pq);
+
+    /* put 8x8 block with bicubic interpolation and quarterpel precision
+     * last argument is actually round value instead of height
+     */
+    op_pixels_func put_vc1_mspel_pixels_tab[16];
+    op_pixels_func avg_vc1_mspel_pixels_tab[16];
+
+    /* This is really one func used in VC-1 decoding */
+    h264_chroma_mc_func put_no_rnd_vc1_chroma_pixels_tab[3];
+    h264_chroma_mc_func avg_no_rnd_vc1_chroma_pixels_tab[3];
+} VC1DSPContext;
+
+void ff_vc1dsp_init(VC1DSPContext* c);
+void ff_vc1dsp_init_altivec(VC1DSPContext* c);
+void ff_vc1dsp_init_mmx(VC1DSPContext* dsp);
+
+#endif /* AVCODEC_VC1DSP_H */
diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c
index 39bf3f2..16ad965 100644
--- a/libavcodec/x86/dsputil_mmx.c
+++ b/libavcodec/x86/dsputil_mmx.c
@@ -1894,20 +1894,14 @@ PREFETCH(prefetch_3dnow, prefetch)
 
 void ff_put_h264_chroma_mc8_mmx_rnd   (uint8_t *dst, uint8_t *src,
                                        int stride, int h, int x, int y);
-void ff_put_vc1_chroma_mc8_mmx_nornd  (uint8_t *dst, uint8_t *src,
-                                       int stride, int h, int x, int y);
 void ff_put_rv40_chroma_mc8_mmx       (uint8_t *dst, uint8_t *src,
                                        int stride, int h, int x, int y);
 void ff_avg_h264_chroma_mc8_mmx2_rnd  (uint8_t *dst, uint8_t *src,
                                        int stride, int h, int x, int y);
-void ff_avg_vc1_chroma_mc8_mmx2_nornd (uint8_t *dst, uint8_t *src,
-                                       int stride, int h, int x, int y);
 void ff_avg_rv40_chroma_mc8_mmx2      (uint8_t *dst, uint8_t *src,
                                        int stride, int h, int x, int y);
 void ff_avg_h264_chroma_mc8_3dnow_rnd (uint8_t *dst, uint8_t *src,
                                        int stride, int h, int x, int y);
-void ff_avg_vc1_chroma_mc8_3dnow_nornd(uint8_t *dst, uint8_t *src,
-                                       int stride, int h, int x, int y);
 void ff_avg_rv40_chroma_mc8_3dnow     (uint8_t *dst, uint8_t *src,
                                        int stride, int h, int x, int y);
 
@@ -1931,15 +1925,11 @@ void ff_avg_h264_chroma_mc2_mmx2      (uint8_t *dst, uint8_t *src,
 
 void ff_put_h264_chroma_mc8_ssse3_rnd (uint8_t *dst, uint8_t *src,
                                        int stride, int h, int x, int y);
-void ff_put_vc1_chroma_mc8_ssse3_nornd(uint8_t *dst, uint8_t *src,
-                                       int stride, int h, int x, int y);
 void ff_put_h264_chroma_mc4_ssse3     (uint8_t *dst, uint8_t *src,
                                        int stride, int h, int x, int y);
 
 void ff_avg_h264_chroma_mc8_ssse3_rnd (uint8_t *dst, uint8_t *src,
                                        int stride, int h, int x, int y);
-void ff_avg_vc1_chroma_mc8_ssse3_nornd(uint8_t *dst, uint8_t *src,
-                                       int stride, int h, int x, int y);
 void ff_avg_h264_chroma_mc4_ssse3     (uint8_t *dst, uint8_t *src,
                                        int stride, int h, int x, int y);
 
@@ -2535,7 +2525,6 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
 #if HAVE_YASM
         c->put_h264_chroma_pixels_tab[0]= ff_put_h264_chroma_mc8_mmx_rnd;
         c->put_h264_chroma_pixels_tab[1]= ff_put_h264_chroma_mc4_mmx;
-        c->put_no_rnd_vc1_chroma_pixels_tab[0]= ff_put_vc1_chroma_mc8_mmx_nornd;
 
         c->put_rv40_chroma_pixels_tab[0]= ff_put_rv40_chroma_mc8_mmx;
         c->put_rv40_chroma_pixels_tab[1]= ff_put_rv40_chroma_mc4_mmx;
@@ -2622,8 +2611,6 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
             c->avg_rv40_chroma_pixels_tab[0]= ff_avg_rv40_chroma_mc8_mmx2;
             c->avg_rv40_chroma_pixels_tab[1]= ff_avg_rv40_chroma_mc4_mmx2;
 
-            c->avg_no_rnd_vc1_chroma_pixels_tab[0]= ff_avg_vc1_chroma_mc8_mmx2_nornd;
-
             c->avg_h264_chroma_pixels_tab[0]= ff_avg_h264_chroma_mc8_mmx2_rnd;
             c->avg_h264_chroma_pixels_tab[1]= ff_avg_h264_chroma_mc4_mmx2;
             c->avg_h264_chroma_pixels_tab[2]= ff_avg_h264_chroma_mc2_mmx2;
@@ -2636,9 +2623,6 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
                 c->add_hfyu_median_prediction = add_hfyu_median_prediction_cmov;
 #endif
 
-            if (CONFIG_VC1_DECODER)
-                ff_vc1dsp_init_mmx(c, avctx);
-
             c->add_png_paeth_prediction= add_png_paeth_prediction_mmx2;
         } else if (mm_flags & AV_CPU_FLAG_3DNOW) {
             c->prefetch = prefetch_3dnow;
@@ -2695,8 +2679,6 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
             c->avg_h264_chroma_pixels_tab[0]= ff_avg_h264_chroma_mc8_3dnow_rnd;
             c->avg_h264_chroma_pixels_tab[1]= ff_avg_h264_chroma_mc4_3dnow;
 
-            c->avg_no_rnd_vc1_chroma_pixels_tab[0]= ff_avg_vc1_chroma_mc8_3dnow_nornd;
-
             c->avg_rv40_chroma_pixels_tab[0]= ff_avg_rv40_chroma_mc8_3dnow;
             c->avg_rv40_chroma_pixels_tab[1]= ff_avg_rv40_chroma_mc4_3dnow;
 #endif
@@ -2745,8 +2727,6 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
             H264_QPEL_FUNCS(3, 3, ssse3);
             c->add_png_paeth_prediction= add_png_paeth_prediction_ssse3;
 #if HAVE_YASM
-            c->put_no_rnd_vc1_chroma_pixels_tab[0]= ff_put_vc1_chroma_mc8_ssse3_nornd;
-            c->avg_no_rnd_vc1_chroma_pixels_tab[0]= ff_avg_vc1_chroma_mc8_ssse3_nornd;
             c->put_h264_chroma_pixels_tab[0]= ff_put_h264_chroma_mc8_ssse3_rnd;
             c->avg_h264_chroma_pixels_tab[0]= ff_avg_h264_chroma_mc8_ssse3_rnd;
             c->put_h264_chroma_pixels_tab[1]= ff_put_h264_chroma_mc4_ssse3;
diff --git a/libavcodec/x86/dsputil_mmx.h b/libavcodec/x86/dsputil_mmx.h
index a095e1e..59a9613 100644
--- a/libavcodec/x86/dsputil_mmx.h
+++ b/libavcodec/x86/dsputil_mmx.h
@@ -196,7 +196,6 @@ void ff_avg_cavs_qpel8_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride);
 void ff_put_cavs_qpel16_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride);
 void ff_avg_cavs_qpel16_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride);
 
-void ff_vc1dsp_init_mmx(DSPContext* dsp, AVCodecContext *avctx);
 void ff_put_vc1_mspel_mc00_mmx(uint8_t *dst, const uint8_t *src, int stride, int rnd);
 void ff_avg_vc1_mspel_mc00_mmx2(uint8_t *dst, const uint8_t *src, int stride, int rnd);
 
diff --git a/libavcodec/x86/vc1dsp_mmx.c b/libavcodec/x86/vc1dsp_mmx.c
index 04b4aba..a525aee 100644
--- a/libavcodec/x86/vc1dsp_mmx.c
+++ b/libavcodec/x86/vc1dsp_mmx.c
@@ -28,6 +28,7 @@
 #include "libavutil/x86_cpu.h"
 #include "libavcodec/dsputil.h"
 #include "dsputil_mmx.h"
+#include "libavcodec/vc1dsp.h"
 
 #define OP_PUT(S,D)
 #define OP_AVG(S,D) "pavgb " #S ", " #D " \n\t"
@@ -712,30 +713,45 @@ static void vc1_h_loop_filter16_sse4(uint8_t *src, int stride, int pq)
     ff_vc1_h_loop_filter8_sse4(src,          stride, pq);
     ff_vc1_h_loop_filter8_sse4(src+8*stride, stride, pq);
 }
+
 #endif
 
-void ff_vc1dsp_init_mmx(DSPContext* dsp, AVCodecContext *avctx) {
+void ff_put_vc1_chroma_mc8_mmx_nornd  (uint8_t *dst, uint8_t *src,
+                                       int stride, int h, int x, int y);
+void ff_avg_vc1_chroma_mc8_mmx2_nornd (uint8_t *dst, uint8_t *src,
+                                       int stride, int h, int x, int y);
+void ff_avg_vc1_chroma_mc8_3dnow_nornd(uint8_t *dst, uint8_t *src,
+                                       int stride, int h, int x, int y);
+void ff_put_vc1_chroma_mc8_ssse3_nornd(uint8_t *dst, uint8_t *src,
+                                       int stride, int h, int x, int y);
+void ff_avg_vc1_chroma_mc8_ssse3_nornd(uint8_t *dst, uint8_t *src,
+                                       int stride, int h, int x, int y);
+
+void ff_vc1dsp_init_mmx(VC1DSPContext *dsp)
+{
     int mm_flags = av_get_cpu_flags();
 
-    dsp->put_vc1_mspel_pixels_tab[ 0] = ff_put_vc1_mspel_mc00_mmx;
-    dsp->put_vc1_mspel_pixels_tab[ 4] = put_vc1_mspel_mc01_mmx;
-    dsp->put_vc1_mspel_pixels_tab[ 8] = put_vc1_mspel_mc02_mmx;
-    dsp->put_vc1_mspel_pixels_tab[12] = put_vc1_mspel_mc03_mmx;
-
-    dsp->put_vc1_mspel_pixels_tab[ 1] = put_vc1_mspel_mc10_mmx;
-    dsp->put_vc1_mspel_pixels_tab[ 5] = put_vc1_mspel_mc11_mmx;
-    dsp->put_vc1_mspel_pixels_tab[ 9] = put_vc1_mspel_mc12_mmx;
-    dsp->put_vc1_mspel_pixels_tab[13] = put_vc1_mspel_mc13_mmx;
-
-    dsp->put_vc1_mspel_pixels_tab[ 2] = put_vc1_mspel_mc20_mmx;
-    dsp->put_vc1_mspel_pixels_tab[ 6] = put_vc1_mspel_mc21_mmx;
-    dsp->put_vc1_mspel_pixels_tab[10] = put_vc1_mspel_mc22_mmx;
-    dsp->put_vc1_mspel_pixels_tab[14] = put_vc1_mspel_mc23_mmx;
-
-    dsp->put_vc1_mspel_pixels_tab[ 3] = put_vc1_mspel_mc30_mmx;
-    dsp->put_vc1_mspel_pixels_tab[ 7] = put_vc1_mspel_mc31_mmx;
-    dsp->put_vc1_mspel_pixels_tab[11] = put_vc1_mspel_mc32_mmx;
-    dsp->put_vc1_mspel_pixels_tab[15] = put_vc1_mspel_mc33_mmx;
+    if (mm_flags & AV_CPU_FLAG_MMX) {
+        dsp->put_vc1_mspel_pixels_tab[ 0] = ff_put_vc1_mspel_mc00_mmx;
+        dsp->put_vc1_mspel_pixels_tab[ 4] = put_vc1_mspel_mc01_mmx;
+        dsp->put_vc1_mspel_pixels_tab[ 8] = put_vc1_mspel_mc02_mmx;
+        dsp->put_vc1_mspel_pixels_tab[12] = put_vc1_mspel_mc03_mmx;
+
+        dsp->put_vc1_mspel_pixels_tab[ 1] = put_vc1_mspel_mc10_mmx;
+        dsp->put_vc1_mspel_pixels_tab[ 5] = put_vc1_mspel_mc11_mmx;
+        dsp->put_vc1_mspel_pixels_tab[ 9] = put_vc1_mspel_mc12_mmx;
+        dsp->put_vc1_mspel_pixels_tab[13] = put_vc1_mspel_mc13_mmx;
+
+        dsp->put_vc1_mspel_pixels_tab[ 2] = put_vc1_mspel_mc20_mmx;
+        dsp->put_vc1_mspel_pixels_tab[ 6] = put_vc1_mspel_mc21_mmx;
+        dsp->put_vc1_mspel_pixels_tab[10] = put_vc1_mspel_mc22_mmx;
+        dsp->put_vc1_mspel_pixels_tab[14] = put_vc1_mspel_mc23_mmx;
+
+        dsp->put_vc1_mspel_pixels_tab[ 3] = put_vc1_mspel_mc30_mmx;
+        dsp->put_vc1_mspel_pixels_tab[ 7] = put_vc1_mspel_mc31_mmx;
+        dsp->put_vc1_mspel_pixels_tab[11] = put_vc1_mspel_mc32_mmx;
+        dsp->put_vc1_mspel_pixels_tab[15] = put_vc1_mspel_mc33_mmx;
+    }
 
     if (mm_flags & AV_CPU_FLAG_MMX2){
         dsp->avg_vc1_mspel_pixels_tab[ 0] = ff_avg_vc1_mspel_mc00_mmx2;
@@ -775,11 +791,16 @@ void ff_vc1dsp_init_mmx(DSPContext* dsp, AVCodecContext *avctx) {
 #if HAVE_YASM
     if (mm_flags & AV_CPU_FLAG_MMX) {
         ASSIGN_LF(mmx);
+        dsp->put_no_rnd_vc1_chroma_pixels_tab[0]= ff_put_vc1_chroma_mc8_mmx_nornd;
     }
     return;
     if (mm_flags & AV_CPU_FLAG_MMX2) {
         ASSIGN_LF(mmx2);
+        dsp->avg_no_rnd_vc1_chroma_pixels_tab[0]= ff_avg_vc1_chroma_mc8_mmx2_nornd;
+    } else if (mm_flags & AV_CPU_FLAG_3DNOW) {
+        dsp->avg_no_rnd_vc1_chroma_pixels_tab[0]= ff_avg_vc1_chroma_mc8_3dnow_nornd;
     }
+
     if (mm_flags & AV_CPU_FLAG_SSE2) {
         dsp->vc1_v_loop_filter8  = ff_vc1_v_loop_filter8_sse2;
         dsp->vc1_h_loop_filter8  = ff_vc1_h_loop_filter8_sse2;
@@ -788,6 +809,8 @@ void ff_vc1dsp_init_mmx(DSPContext* dsp, AVCodecContext *avctx) {
     }
     if (mm_flags & AV_CPU_FLAG_SSSE3) {
         ASSIGN_LF(ssse3);
+        dsp->put_no_rnd_vc1_chroma_pixels_tab[0]= ff_put_vc1_chroma_mc8_ssse3_nornd;
+        dsp->avg_no_rnd_vc1_chroma_pixels_tab[0]= ff_avg_vc1_chroma_mc8_ssse3_nornd;
     }
     if (mm_flags & AV_CPU_FLAG_SSE4) {
         dsp->vc1_h_loop_filter8  = ff_vc1_h_loop_filter8_sse4;
-- 
1.7.2.1




More information about the ffmpeg-devel mailing list