cosmetics: reorder mc_luma/mc_chroma/get_ref arguments for consistency with other...

git-svn-id: svn://svn.videolan.org/x264/trunk@694 df754926-b1dd-0310-bc7b-ec298dee348c

diff --git a/common/amd64/mc-a.asm b/common/amd64/mc-a.asm
index cafce5083cde44dc287247814d0377f8d4c84c97..bd53329121d4c3cbd194ac961ef25086b29b61e3 100644 (file)
--- a/common/amd64/mc-a.asm
+++ b/common/amd64/mc-a.asm
@@ -421,8 +421,8 @@ ALIGN 4
 ;=============================================================================
 
 ;-----------------------------------------------------------------------------
-;   void x264_mc_chroma_mmxext( uint8_t *src, int i_src_stride,
-;                               uint8_t *dst, int i_dst_stride,
+;   void x264_mc_chroma_mmxext( uint8_t *dst, int i_dst_stride,
+;                               uint8_t *src, int i_src_stride,
 ;                               int dx, int dy,
 ;                               int i_width, int i_height )
 ;-----------------------------------------------------------------------------
@@ -431,12 +431,12 @@ cglobal x264_mc_chroma_mmxext
     mov     r11d, parm5d
     sar     r10d, 3
     sar     r11d, 3
-    imul    r10d, parm2d
+    imul    r10d, parm4d
     pxor    mm3, mm3
     add     r10d, r11d
     movsxd   r10, r10d
     mov     r11d, parm8d
-    add   parm1q, r10           ; src += (dx>>3) + (dy>>3) * src_stride
+    add   parm3q, r10           ; src += (dx>>3) + (dy>>3) * src_stride
     and   parm5d, 7             ; dx &= 7
     je      .mc1d
     and   parm6d, 7             ; dy &= 7
@@ -460,13 +460,13 @@ cglobal x264_mc_chroma_mmxext
     pmullw  mm6, mm4            ; mm6 = (8-dx)*dy =     cC
     pmullw  mm4, mm0            ; mm4 = (8-dx)*(8-dy) = cA
 
-    mov     rax, parm1q
-    mov     r10, parm3q
+    mov     rax, parm3q
+    mov     r10, parm1q
 
 ALIGN 4
 .height_loop
 
-    movd    mm1, [rax+parm2q]
+    movd    mm1, [rax+parm4q]
     movd    mm0, [rax]
     punpcklbw mm1, mm3          ; 00 px1 | 00 px2 | 00 px3 | 00 px4
     punpcklbw mm0, mm3
@@ -476,7 +476,7 @@ ALIGN 4
     paddw   mm0, mm1            ; mm0 <- result
 
     movd    mm2, [rax+1]
-    movd    mm1, [rax+parm2q+1]
+    movd    mm1, [rax+parm4q+1]
     punpcklbw mm2, mm3
     punpcklbw mm1, mm3
 
@@ -491,16 +491,16 @@ ALIGN 4
     packuswb mm0, mm3           ; 00 00 00 00 px1 px2 px3 px4
     movd    [r10], mm0
 
-    add     rax, parm2q
-    add     r10, parm4q         ; i_dst_stride
+    add     rax, parm4q
+    add     r10, parm2q         ; i_dst_stride
     dec     r11d
     jnz     .height_loop
 
     sub     parm7d, 8
     jnz     .finish             ; width != 8 so assume 4
 
-    mov     r10, parm3q         ; dst
-    mov     rax, parm1q         ; src
+    mov     r10, parm1q         ; dst
+    mov     rax, parm3q         ; src
     mov     r11d, parm8d        ; i_height
     add     r10, 4
     add     rax, 4
@@ -511,17 +511,13 @@ ALIGN 4
 
 ALIGN 4
 .mc1d
-%ifdef WIN64
-%define pel_offset rsi
-%else
-%define pel_offset r9
-%endif
+%define pel_offset temp1q
     mov       eax, parm5d
     or        eax, parm6d
     and       eax, 7
     cmp    parm5d, 0
     mov    pel_offset, 1
-    cmove  pel_offset, parm2q   ; pel_offset = dx ? 1 : src_stride
+    cmove  pel_offset, parm4q   ; pel_offset = dx ? 1 : src_stride
     movd      mm6, eax
     movq      mm5, [pw_8 GLOBAL]
     pshufw    mm6, mm6, 0
@@ -533,8 +529,8 @@ ALIGN 4
 
 ALIGN 4
 .height_loop1_w4
-    movd      mm0, [parm1q+pel_offset]
-    movd      mm1, [parm1q]
+    movd      mm0, [parm3q+pel_offset]
+    movd      mm1, [parm3q]
     punpcklbw mm0, mm3
     punpcklbw mm1, mm3
     pmullw    mm0, mm6
@@ -543,17 +539,17 @@ ALIGN 4
     paddw     mm0, mm1
     psrlw     mm0, 3
     packuswb  mm0, mm3
-    movd [parm3q], mm0
-    add    parm1q, parm2q
+    movd [parm1q], mm0
     add    parm3q, parm4q
+    add    parm1q, parm2q
     dec      r11d
     jnz .height_loop1_w4
     rep ret
 
 ALIGN 4
 .height_loop1_w8
-    movq      mm0, [parm1q+pel_offset]
-    movq      mm1, [parm1q]
+    movq      mm0, [parm3q+pel_offset]
+    movq      mm1, [parm3q]
     movq      mm2, mm0
     movq      mm4, mm1
     punpcklbw mm0, mm3
@@ -571,9 +567,9 @@ ALIGN 4
     psrlw     mm0, 3
     psrlw     mm2, 3
     packuswb  mm0, mm2
-    movq [parm3q], mm0
-    add    parm1q, parm2q
+    movq [parm1q], mm0
     add    parm3q, parm4q
+    add    parm1q, parm2q
     dec      r11d
     jnz .height_loop1_w8
     rep ret

diff --git a/common/i386/mc-a.asm b/common/i386/mc-a.asm
index 719107391d0a7a5425aeadd4f7c87041f782d17b..991b09665042197386d4244b1b208a73bb54d634 100644 (file)
--- a/common/i386/mc-a.asm
+++ b/common/i386/mc-a.asm
@@ -484,8 +484,8 @@ ALIGN 4
 ;=============================================================================
 
 ;-----------------------------------------------------------------------------
-;   void x264_mc_chroma_mmxext( uint8_t *src, int i_src_stride,
-;                               uint8_t *dst, int i_dst_stride,
+;   void x264_mc_chroma_mmxext( uint8_t *dst, int i_dst_stride,
+;                               uint8_t *src, int i_src_stride,
 ;                               int dx, int dy,
 ;                               int i_width, int i_height )
 ;-----------------------------------------------------------------------------
@@ -501,9 +501,9 @@ cglobal x264_mc_chroma_mmxext
     mov     edi, edx
     sar     ecx, 3
     sar     edx, 3
-    imul    ecx, [picesp+4+8]
+    imul    ecx, [picesp+4+16]
     add     ecx, edx
-    add     [picesp+4+4], ecx     ; src += (dx>>3) + (dy>>3) * src_stride
+    add     [picesp+4+12], ecx   ; src += (dx>>3) + (dy>>3) * src_stride
 
     pxor    mm3, mm3
 
@@ -526,9 +526,9 @@ cglobal x264_mc_chroma_mmxext
     pmullw  mm6, mm4            ; mm6 = (8-dx)*dy =     cC
     pmullw  mm4, mm0            ; mm4 = (8-dx)*(8-dy) = cA
 
-    mov     eax, [picesp+4+4]   ; src
-    mov     edi, [picesp+4+12]  ; dst
-    mov     ecx, [picesp+4+8]   ; i_src_stride
+    mov     eax, [picesp+4+12]  ; src
+    mov     edi, [picesp+4+4]   ; dst
+    mov     ecx, [picesp+4+16]  ; i_src_stride
     mov     edx, [picesp+4+32]  ; i_height
 
 ALIGN 4
@@ -560,7 +560,7 @@ ALIGN 4
     movd    [edi], mm0
 
     add     eax, ecx
-    add     edi, [picesp+4+16]
+    add     edi, [picesp+4+8]
 
     dec     edx
     jnz     .height_loop
@@ -568,8 +568,8 @@ ALIGN 4
     sub     [picesp+4+28], dword 8
     jnz     .finish            ; width != 8 so assume 4
 
-    mov     edi, [picesp+4+12] ; dst
-    mov     eax, [picesp+4+4]  ; src
+    mov     edi, [picesp+4+4]  ; dst
+    mov     eax, [picesp+4+12] ; src
     mov     edx, [picesp+4+32] ; i_height
     add     edi, 4
     add     eax, 4

diff --git a/common/i386/mc-c.c b/common/i386/mc-c.c
index e60c15f3ceef12489bece310e554cdb36daec93b..8ac3bf66cf6c803ae06b333c7b07b70c5850ff33 100644 (file)
--- a/common/i386/mc-c.c
+++ b/common/i386/mc-c.c
@@ -84,8 +84,8 @@ static void (* const x264_mc_copy_wtab_mmx[5])( uint8_t *, int, uint8_t *, int,
 static const int hpel_ref0[16] = {0,1,1,1,0,1,1,1,2,3,3,3,0,1,1,1};
 static const int hpel_ref1[16] = {0,0,0,0,2,2,3,2,2,2,3,2,2,2,3,2};
 
-void mc_luma_mmx( uint8_t *src[4], int i_src_stride,
-                  uint8_t *dst,    int i_dst_stride,
+void mc_luma_mmx( uint8_t *dst,    int i_dst_stride,
+                  uint8_t *src[4], int i_src_stride,
                   int mvx, int mvy,
                   int i_width, int i_height )
 {
@@ -107,8 +107,8 @@ void mc_luma_mmx( uint8_t *src[4], int i_src_stride,
     }
 }
 
-uint8_t *get_ref_mmx( uint8_t *src[4], int i_src_stride,
-                      uint8_t *dst,   int *i_dst_stride,
+uint8_t *get_ref_mmx( uint8_t *dst,   int *i_dst_stride,
+                      uint8_t *src[4], int i_src_stride,
                       int mvx, int mvy,
                       int i_width, int i_height )
 {

diff --git a/common/macroblock.c b/common/macroblock.c
index 6bd2c83b0bff462f53eefc4502892dda56bad42c..86c82dbb2db514b6d0ce8ce7594bd7cd80a01148 100644 (file)
--- a/common/macroblock.c
+++ b/common/macroblock.c
@@ -589,21 +589,21 @@ static inline void x264_mb_mc_0xywh( x264_t *h, int x, int y, int width, int hei
     const int mvx   = x264_clip3( h->mb.cache.mv[0][i8][0], h->mb.mv_min[0], h->mb.mv_max[0] );
     int       mvy   = x264_clip3( h->mb.cache.mv[0][i8][1], h->mb.mv_min[1], h->mb.mv_max[1] );
 
-    h->mc.mc_luma( h->mb.pic.p_fref[0][i_ref], h->mb.pic.i_stride[0],
-                    &h->mb.pic.p_fdec[0][4*y*FDEC_STRIDE+4*x], FDEC_STRIDE,
-                    mvx + 4*4*x, mvy + 4*4*y, 4*width, 4*height );
+    h->mc.mc_luma( &h->mb.pic.p_fdec[0][4*y*FDEC_STRIDE+4*x], FDEC_STRIDE,
+                   h->mb.pic.p_fref[0][i_ref], h->mb.pic.i_stride[0],
+                   mvx + 4*4*x, mvy + 4*4*y, 4*width, 4*height );
 
     // chroma is offset if MCing from a field of opposite parity
     if( h->mb.b_interlaced & i_ref )
         mvy += (h->mb.i_mb_y & 1)*4 - 2;
 
-    h->mc.mc_chroma( &h->mb.pic.p_fref[0][i_ref][4][2*y*h->mb.pic.i_stride[1]+2*x], h->mb.pic.i_stride[1],
-                      &h->mb.pic.p_fdec[1][2*y*FDEC_STRIDE+2*x], FDEC_STRIDE,
-                      mvx, mvy, 2*width, 2*height );
+    h->mc.mc_chroma( &h->mb.pic.p_fdec[1][2*y*FDEC_STRIDE+2*x], FDEC_STRIDE,
+                     &h->mb.pic.p_fref[0][i_ref][4][2*y*h->mb.pic.i_stride[1]+2*x], h->mb.pic.i_stride[1],
+                     mvx, mvy, 2*width, 2*height );
 
-    h->mc.mc_chroma( &h->mb.pic.p_fref[0][i_ref][5][2*y*h->mb.pic.i_stride[2]+2*x], h->mb.pic.i_stride[2],
-                      &h->mb.pic.p_fdec[2][2*y*FDEC_STRIDE+2*x], FDEC_STRIDE,
-                      mvx, mvy, 2*width, 2*height );
+    h->mc.mc_chroma( &h->mb.pic.p_fdec[2][2*y*FDEC_STRIDE+2*x], FDEC_STRIDE,
+                     &h->mb.pic.p_fref[0][i_ref][5][2*y*h->mb.pic.i_stride[2]+2*x], h->mb.pic.i_stride[2],
+                     mvx, mvy, 2*width, 2*height );
 }
 static inline void x264_mb_mc_1xywh( x264_t *h, int x, int y, int width, int height )
 {
@@ -612,20 +612,20 @@ static inline void x264_mb_mc_1xywh( x264_t *h, int x, int y, int width, int hei
     const int mvx   = x264_clip3( h->mb.cache.mv[1][i8][0], h->mb.mv_min[0], h->mb.mv_max[0] );
     int       mvy   = x264_clip3( h->mb.cache.mv[1][i8][1], h->mb.mv_min[1], h->mb.mv_max[1] );
 
-    h->mc.mc_luma( h->mb.pic.p_fref[1][i_ref], h->mb.pic.i_stride[0],
-                    &h->mb.pic.p_fdec[0][4*y*FDEC_STRIDE+4*x], FDEC_STRIDE,
-                    mvx + 4*4*x, mvy + 4*4*y, 4*width, 4*height );
+    h->mc.mc_luma( &h->mb.pic.p_fdec[0][4*y*FDEC_STRIDE+4*x], FDEC_STRIDE,
+                   h->mb.pic.p_fref[1][i_ref], h->mb.pic.i_stride[0],
+                   mvx + 4*4*x, mvy + 4*4*y, 4*width, 4*height );
 
     if( h->mb.b_interlaced & i_ref )
         mvy += (h->mb.i_mb_y & 1)*4 - 2;
 
-    h->mc.mc_chroma( &h->mb.pic.p_fref[1][i_ref][4][2*y*h->mb.pic.i_stride[1]+2*x], h->mb.pic.i_stride[1],
-                      &h->mb.pic.p_fdec[1][2*y*FDEC_STRIDE+2*x], FDEC_STRIDE,
-                      mvx, mvy, 2*width, 2*height );
+    h->mc.mc_chroma( &h->mb.pic.p_fdec[1][2*y*FDEC_STRIDE+2*x], FDEC_STRIDE,
+                     &h->mb.pic.p_fref[1][i_ref][4][2*y*h->mb.pic.i_stride[1]+2*x], h->mb.pic.i_stride[1],
+                     mvx, mvy, 2*width, 2*height );
 
-    h->mc.mc_chroma( &h->mb.pic.p_fref[1][i_ref][5][2*y*h->mb.pic.i_stride[2]+2*x], h->mb.pic.i_stride[2],
-                      &h->mb.pic.p_fdec[2][2*y*FDEC_STRIDE+2*x], FDEC_STRIDE,
-                      mvx, mvy, 2*width, 2*height );
+    h->mc.mc_chroma( &h->mb.pic.p_fdec[2][2*y*FDEC_STRIDE+2*x], FDEC_STRIDE,
+                     &h->mb.pic.p_fref[1][i_ref][5][2*y*h->mb.pic.i_stride[2]+2*x], h->mb.pic.i_stride[2],
+                     mvx, mvy, 2*width, 2*height );
 }
 
 static inline void x264_mb_mc_01xywh( x264_t *h, int x, int y, int width, int height )
@@ -640,8 +640,8 @@ static inline void x264_mb_mc_01xywh( x264_t *h, int x, int y, int width, int he
 
     x264_mb_mc_0xywh( h, x, y, width, height );
 
-    h->mc.mc_luma( h->mb.pic.p_fref[1][i_ref1], h->mb.pic.i_stride[0],
-                    tmp, 16, mvx1 + 4*4*x, mvy1 + 4*4*y, 4*width, 4*height );
+    h->mc.mc_luma( tmp, 16, h->mb.pic.p_fref[1][i_ref1], h->mb.pic.i_stride[0],
+                   mvx1 + 4*4*x, mvy1 + 4*4*y, 4*width, 4*height );
 
     if( h->mb.b_interlaced & i_ref1 )
         mvy1 += (h->mb.i_mb_y & 1)*4 - 2;
@@ -653,24 +653,24 @@ static inline void x264_mb_mc_01xywh( x264_t *h, int x, int y, int width, int he
 
         h->mc.avg_weight[i_mode]( &h->mb.pic.p_fdec[0][4*y*FDEC_STRIDE+4*x], FDEC_STRIDE, tmp, 16, weight );
 
-        h->mc.mc_chroma( &h->mb.pic.p_fref[1][i_ref1][4][2*y*h->mb.pic.i_stride[1]+2*x], h->mb.pic.i_stride[1],
-                          tmp, 16, mvx1, mvy1, 2*width, 2*height );
+        h->mc.mc_chroma( tmp, 16, &h->mb.pic.p_fref[1][i_ref1][4][2*y*h->mb.pic.i_stride[1]+2*x], h->mb.pic.i_stride[1],
+                         mvx1, mvy1, 2*width, 2*height );
         h->mc.avg_weight[i_mode+3]( &h->mb.pic.p_fdec[1][2*y*FDEC_STRIDE+2*x], FDEC_STRIDE, tmp, 16, weight );
 
-        h->mc.mc_chroma( &h->mb.pic.p_fref[1][i_ref1][5][2*y*h->mb.pic.i_stride[2]+2*x], h->mb.pic.i_stride[2],
-                          tmp, 16, mvx1, mvy1, 2*width, 2*height );
+        h->mc.mc_chroma( tmp, 16, &h->mb.pic.p_fref[1][i_ref1][5][2*y*h->mb.pic.i_stride[2]+2*x], h->mb.pic.i_stride[2],
+                         mvx1, mvy1, 2*width, 2*height );
         h->mc.avg_weight[i_mode+3]( &h->mb.pic.p_fdec[2][2*y*FDEC_STRIDE+2*x], FDEC_STRIDE, tmp, 16, weight );
     }
     else
     {
         h->mc.avg[i_mode]( &h->mb.pic.p_fdec[0][4*y*FDEC_STRIDE+4*x], FDEC_STRIDE, tmp, 16 );
 
-        h->mc.mc_chroma( &h->mb.pic.p_fref[1][i_ref1][4][2*y*h->mb.pic.i_stride[1]+2*x], h->mb.pic.i_stride[1],
-                          tmp, 16, mvx1, mvy1, 2*width, 2*height );
+        h->mc.mc_chroma( tmp, 16, &h->mb.pic.p_fref[1][i_ref1][4][2*y*h->mb.pic.i_stride[1]+2*x], h->mb.pic.i_stride[1],
+                         mvx1, mvy1, 2*width, 2*height );
         h->mc.avg[i_mode+3]( &h->mb.pic.p_fdec[1][2*y*FDEC_STRIDE+2*x], FDEC_STRIDE, tmp, 16 );
 
-        h->mc.mc_chroma( &h->mb.pic.p_fref[1][i_ref1][5][2*y*h->mb.pic.i_stride[2]+2*x], h->mb.pic.i_stride[2],
-                          tmp, 16, mvx1, mvy1, 2*width, 2*height );
+        h->mc.mc_chroma( tmp, 16, &h->mb.pic.p_fref[1][i_ref1][5][2*y*h->mb.pic.i_stride[2]+2*x], h->mb.pic.i_stride[2],
+                         mvx1, mvy1, 2*width, 2*height );
         h->mc.avg[i_mode+3]( &h->mb.pic.p_fdec[2][2*y*FDEC_STRIDE+2*x], FDEC_STRIDE, tmp, 16 );
     }
 }

diff --git a/common/mc.c b/common/mc.c
index a88e884583069e6887c1623608e5b4032756992a..00547adbdfcb83a17d7020952ee44180b0ff7090 100644 (file)
--- a/common/mc.c
+++ b/common/mc.c
@@ -231,8 +231,8 @@ static void hpel_filter( uint8_t *dsth, uint8_t *dstv, uint8_t *dstc, uint8_t *s
 static const int hpel_ref0[16] = {0,1,1,1,0,1,1,1,2,3,3,3,0,1,1,1};
 static const int hpel_ref1[16] = {0,0,0,0,2,2,3,2,2,2,3,2,2,2,3,2};
 
-static void mc_luma( uint8_t *src[4], int i_src_stride,
-                     uint8_t *dst,    int i_dst_stride,
+static void mc_luma( uint8_t *dst,    int i_dst_stride,
+                     uint8_t *src[4], int i_src_stride,
                      int mvx, int mvy,
                      int i_width, int i_height )
 {
@@ -252,8 +252,8 @@ static void mc_luma( uint8_t *src[4], int i_src_stride,
     }
 }
 
-static uint8_t *get_ref( uint8_t *src[4], int i_src_stride,
-                         uint8_t *dst,   int *i_dst_stride,
+static uint8_t *get_ref( uint8_t *dst,   int *i_dst_stride,
+                         uint8_t *src[4], int i_src_stride,
                          int mvx, int mvy,
                          int i_width, int i_height )
 {
@@ -276,10 +276,10 @@ static uint8_t *get_ref( uint8_t *src[4], int i_src_stride,
 }
 
 /* full chroma mc (ie until 1/8 pixel)*/
-static void motion_compensation_chroma( uint8_t *src, int i_src_stride,
-                                        uint8_t *dst, int i_dst_stride,
-                                        int mvx, int mvy,
-                                        int i_width, int i_height )
+static void mc_chroma( uint8_t *dst, int i_dst_stride,
+                       uint8_t *src, int i_src_stride,
+                       int mvx, int mvy,
+                       int i_width, int i_height )
 {
     uint8_t *srcp;
     int x, y;
@@ -340,7 +340,7 @@ void x264_mc_init( int cpu, x264_mc_functions_t *pf )
 {
     pf->mc_luma   = mc_luma;
     pf->get_ref   = get_ref;
-    pf->mc_chroma = motion_compensation_chroma;
+    pf->mc_chroma = mc_chroma;
 
     pf->avg[PIXEL_16x16]= pixel_avg_16x16;
     pf->avg[PIXEL_16x8] = pixel_avg_16x8;

diff --git a/common/mc.h b/common/mc.h
index 80db9d42e5b87c90ca7757fb2bdb5e89e72e4185..1a2f64f92992118699fcb28b5849085271e9b59c 100644 (file)
--- a/common/mc.h
+++ b/common/mc.h
@@ -33,18 +33,18 @@
 
 typedef struct
 {
-    void (*mc_luma)(uint8_t **, int, uint8_t *, int,
+    void (*mc_luma)(uint8_t *dst, int i_dst, uint8_t **src, int i_src,
                     int mvx, int mvy,
                     int i_width, int i_height );
 
     /* may round up the dimensions if they're not a power of 2 */
-    uint8_t* (*get_ref)(uint8_t **, int, uint8_t *, int *,
+    uint8_t* (*get_ref)(uint8_t *dst, int *i_dst, uint8_t **src, int i_src,
                         int mvx, int mvy,
                         int i_width, int i_height );
 
     /* mc_chroma may write up to 2 bytes of garbage to the right of dst,
      * so it must be run from left to right. */
-    void (*mc_chroma)(uint8_t *, int, uint8_t *, int,
+    void (*mc_chroma)(uint8_t *dst, int i_dst, uint8_t *src, int i_src,
                       int mvx, int mvy,
                       int i_width, int i_height );
 

diff --git a/common/ppc/mc.c b/common/ppc/mc.c
index bd6c514a33fba7dd03483550d01dacf76bab147f..bd524c27636eac050742e65de9b344500c500287 100644 (file)
--- a/common/ppc/mc.c
+++ b/common/ppc/mc.c
@@ -132,8 +132,8 @@ MC_COPY( mc_copy_w4,  4  )
 MC_COPY( mc_copy_w8,  8  )
 MC_COPY( mc_copy_w16, 16 )
 
-void mc_luma_altivec( uint8_t *src[4], int i_src_stride,
-                      uint8_t *dst,    int i_dst_stride,
+void mc_luma_altivec( uint8_t *dst,    int i_dst_stride,
+                      uint8_t *src[4], int i_src_stride,
                       int mvx, int mvy,
                       int i_width, int i_height )
 {
@@ -190,8 +190,8 @@ void mc_luma_altivec( uint8_t *src[4], int i_src_stride,
     }
 }
 
-uint8_t *get_ref_altivec( uint8_t *src[4], int i_src_stride,
-                          uint8_t *dst,    int * i_dst_stride,
+uint8_t *get_ref_altivec( uint8_t *dst,    int * i_dst_stride,
+                          uint8_t *src[4], int i_src_stride,
                           int mvx, int mvy,
                           int i_width, int i_height )
 {
@@ -253,8 +253,8 @@ uint8_t *get_ref_altivec( uint8_t *src[4], int i_src_stride,
         src##a##v_16 = vec_mladd( coeff##a##v, src##a##v_16, zero_u16v ); \
         dstv_16      = vec_add( dstv_16, src##a##v_16 )
 
-static void mc_chroma_altivec_4xh( uint8_t *src, int i_src_stride,
-                                   uint8_t *dst, int i_dst_stride,
+static void mc_chroma_altivec_4xh( uint8_t *dst, int i_dst_stride,
+                                   uint8_t *src, int i_src_stride,
                                    int mvx, int mvy,
                                    int i_height )
 {
@@ -319,8 +319,8 @@ static void mc_chroma_altivec_4xh( uint8_t *src, int i_src_stride,
     }
 }
 
-static void mc_chroma_altivec_8xh( uint8_t *src, int i_src_stride,
-                                   uint8_t *dst, int i_dst_stride,
+static void mc_chroma_altivec_8xh( uint8_t *dst, int i_dst_stride,
+                                   uint8_t *src, int i_src_stride,
                                    int mvx, int mvy,
                                    int i_height )
 {
@@ -385,19 +385,19 @@ static void mc_chroma_altivec_8xh( uint8_t *src, int i_src_stride,
     }
 }
 
-static void mc_chroma_altivec( uint8_t *src, int i_src_stride,
-                               uint8_t *dst, int i_dst_stride,
+static void mc_chroma_altivec( uint8_t *dst, int i_dst_stride,
+                               uint8_t *src, int i_src_stride,
                                int mvx, int mvy,
                                int i_width, int i_height )
 {
     if( i_width == 8 )
     {
-        mc_chroma_altivec_8xh( src, i_src_stride, dst, i_dst_stride,
+        mc_chroma_altivec_8xh( dst, i_dst_stride, src, i_src_stride,
                                mvx, mvy, i_height );
     }
     else
     {
-        mc_chroma_altivec_4xh( src, i_src_stride, dst, i_dst_stride,
+        mc_chroma_altivec_4xh( dst, i_dst_stride, src, i_src_stride,
                                mvx, mvy, i_height );
     }
 }

diff --git a/encoder/analyse.c b/encoder/analyse.c
index 1e1db9ea34c4a31671737a33d2d56c52947b6dda..c403d4156aea588896309c6d0a27485ed594ee8c 100644 (file)
--- a/encoder/analyse.c
+++ b/encoder/analyse.c
@@ -1236,8 +1236,8 @@ static int x264_mb_analyse_inter_p4x4_chroma( x264_t *h, x264_mb_analysis_t *a,
     const int oe = 4*(i8x8&1) + 2*(i8x8&2)*FENC_STRIDE;
 
 #define CHROMA4x4MC( width, height, me, x, y ) \
-    h->mc.mc_chroma( &p_fref[4][or+x+y*i_stride], i_stride, &pix1[x+y*16], 16, (me).mv[0], (me).mv[1], width, height ); \
-    h->mc.mc_chroma( &p_fref[5][or+x+y*i_stride], i_stride, &pix2[x+y*16], 16, (me).mv[0], (me).mv[1], width, height );
+    h->mc.mc_chroma( &pix1[x+y*16], 16, &p_fref[4][or+x+y*i_stride], i_stride, (me).mv[0], (me).mv[1], width, height ); \
+    h->mc.mc_chroma( &pix2[x+y*16], 16, &p_fref[5][or+x+y*i_stride], i_stride, (me).mv[0], (me).mv[1], width, height );
 
     if( pixel == PIXEL_4x4 )
     {
@@ -1488,27 +1488,29 @@ static void x264_mb_analyse_inter_b16x16( x264_t *h, x264_mb_analysis_t *a )
     if ( ((a->l0.me16x16.mv[0] | a->l0.me16x16.mv[1]) & 1) == 0 )
     {
         /* l0 reference is halfpel, so get_ref on it will make it faster */
-        src2 = h->mc.get_ref( h->mb.pic.p_fref[0][a->l0.i_ref], h->mb.pic.i_stride[0],
-                        pix2, &stride2,
-                        a->l0.me16x16.mv[0], a->l0.me16x16.mv[1],
-                        16, 16 );
-        h->mc.mc_luma( h->mb.pic.p_fref[1][a->l1.i_ref], h->mb.pic.i_stride[0],
-                        pix1, 16,
-                        a->l1.me16x16.mv[0], a->l1.me16x16.mv[1],
-                        16, 16 );
+        src2 = 
+        h->mc.get_ref( pix2, &stride2,
+                       h->mb.pic.p_fref[0][a->l0.i_ref], h->mb.pic.i_stride[0],
+                       a->l0.me16x16.mv[0], a->l0.me16x16.mv[1],
+                       16, 16 );
+        h->mc.mc_luma( pix1, 16,
+                       h->mb.pic.p_fref[1][a->l1.i_ref], h->mb.pic.i_stride[0],
+                       a->l1.me16x16.mv[0], a->l1.me16x16.mv[1],
+                       16, 16 );
         weight = 64 - weight;
     } 
     else
     {
         /* if l0 was qpel, we'll use get_ref on l1 instead */
-        h->mc.mc_luma( h->mb.pic.p_fref[0][a->l0.i_ref], h->mb.pic.i_stride[0],
-                        pix1, 16,
-                        a->l0.me16x16.mv[0], a->l0.me16x16.mv[1],
-                        16, 16 );
-        src2 = h->mc.get_ref( h->mb.pic.p_fref[1][a->l1.i_ref], h->mb.pic.i_stride[0],
-                        pix2, &stride2,
-                        a->l1.me16x16.mv[0], a->l1.me16x16.mv[1],
-                        16, 16 );
+        h->mc.mc_luma( pix1, 16,
+                       h->mb.pic.p_fref[0][a->l0.i_ref], h->mb.pic.i_stride[0],
+                       a->l0.me16x16.mv[0], a->l0.me16x16.mv[1],
+                       16, 16 );
+        src2 =
+        h->mc.get_ref( pix2, &stride2,
+                       h->mb.pic.p_fref[1][a->l1.i_ref], h->mb.pic.i_stride[0],
+                       a->l1.me16x16.mv[0], a->l1.me16x16.mv[1],
+                       16, 16 );
     }
 
     if( h->param.analyse.b_weighted_bipred )
@@ -1650,8 +1652,8 @@ static void x264_mb_analyse_inter_b8x8( x264_t *h, x264_mb_analysis_t *a )
             x264_macroblock_cache_mv( h, 2*x8, 2*y8, 2, 2, l, m->mv[0], m->mv[1] );
 
             /* BI mode */
-            h->mc.mc_luma( m->p_fref, m->i_stride[0], pix[l], 8,
-                            m->mv[0], m->mv[1], 8, 8 );
+            h->mc.mc_luma( pix[l], 8, m->p_fref, m->i_stride[0],
+                           m->mv[0], m->mv[1], 8, 8 );
             i_part_cost_bi += m->cost_mv;
             /* FIXME: ref cost */
         }
@@ -1715,8 +1717,8 @@ static void x264_mb_analyse_inter_b16x8( x264_t *h, x264_mb_analysis_t *a )
             x264_me_search( h, m, mvc, 2 );
 
             /* BI mode */
-            h->mc.mc_luma( m->p_fref, m->i_stride[0], pix[l], 16,
-                            m->mv[0], m->mv[1], 16, 8 );
+            h->mc.mc_luma( pix[l], 16, m->p_fref, m->i_stride[0],
+                           m->mv[0], m->mv[1], 16, 8 );
             /* FIXME: ref cost */
             i_part_cost_bi += m->cost_mv;
         }
@@ -1785,8 +1787,8 @@ static void x264_mb_analyse_inter_b8x16( x264_t *h, x264_mb_analysis_t *a )
             x264_me_search( h, m, mvc, 2 );
 
             /* BI mode */
-            h->mc.mc_luma( m->p_fref, m->i_stride[0], pix[l], 8,
-                            m->mv[0], m->mv[1], 8, 16 );
+            h->mc.mc_luma( pix[l], 8, m->p_fref, m->i_stride[0],
+                           m->mv[0], m->mv[1], 8, 16 );
             /* FIXME: ref cost */
             i_part_cost_bi += m->cost_mv;
         }

diff --git a/encoder/macroblock.c b/encoder/macroblock.c
index c0967fe0d91b7c8bb660e59d2c28d8dcd5eb8c60..9a96181f0dc25adb17f31df8dbb5a8d79ac88b7c 100644 (file)
--- a/encoder/macroblock.c
+++ b/encoder/macroblock.c
@@ -296,17 +296,17 @@ void x264_macroblock_encode_pskip( x264_t *h )
                                 h->mb.mv_min[1], h->mb.mv_max[1] );
 
     /* Motion compensation XXX probably unneeded */
-    h->mc.mc_luma( h->mb.pic.p_fref[0][0], h->mb.pic.i_stride[0],
-                   h->mb.pic.p_fdec[0],    FDEC_STRIDE,
+    h->mc.mc_luma( h->mb.pic.p_fdec[0],    FDEC_STRIDE,
+                   h->mb.pic.p_fref[0][0], h->mb.pic.i_stride[0],
                    mvx, mvy, 16, 16 );
 
     /* Chroma MC */
-    h->mc.mc_chroma( h->mb.pic.p_fref[0][0][4], h->mb.pic.i_stride[1],
-                     h->mb.pic.p_fdec[1],       FDEC_STRIDE,
+    h->mc.mc_chroma( h->mb.pic.p_fdec[1],       FDEC_STRIDE,
+                     h->mb.pic.p_fref[0][0][4], h->mb.pic.i_stride[1],
                      mvx, mvy, 8, 8 );
 
-    h->mc.mc_chroma( h->mb.pic.p_fref[0][0][5], h->mb.pic.i_stride[2],
-                     h->mb.pic.p_fdec[2],       FDEC_STRIDE,
+    h->mc.mc_chroma( h->mb.pic.p_fdec[2],       FDEC_STRIDE,
+                     h->mb.pic.p_fref[0][0][5], h->mb.pic.i_stride[2],
                      mvx, mvy, 8, 8 );
 
     x264_macroblock_encode_skip( h );
@@ -618,8 +618,8 @@ int x264_macroblock_probe_skip( x264_t *h, const int b_bidir )
         mvp[1] = x264_clip3( h->mb.cache.pskip_mv[1], h->mb.mv_min[1], h->mb.mv_max[1] );
 
         /* Motion compensation */
-        h->mc.mc_luma( h->mb.pic.p_fref[0][0], h->mb.pic.i_stride[0],
-                       h->mb.pic.p_fdec[0],    FDEC_STRIDE,
+        h->mc.mc_luma( h->mb.pic.p_fdec[0],    FDEC_STRIDE,
+                       h->mb.pic.p_fref[0][0], h->mb.pic.i_stride[0],
                        mvp[0], mvp[1], 16, 16 );
     }
 
@@ -657,8 +657,8 @@ int x264_macroblock_probe_skip( x264_t *h, const int b_bidir )
 
         if( !b_bidir )
         {
-            h->mc.mc_chroma( h->mb.pic.p_fref[0][0][4+ch], h->mb.pic.i_stride[1+ch],
-                             h->mb.pic.p_fdec[1+ch],       FDEC_STRIDE,
+            h->mc.mc_chroma( h->mb.pic.p_fdec[1+ch],       FDEC_STRIDE,
+                             h->mb.pic.p_fref[0][0][4+ch], h->mb.pic.i_stride[1+ch],
                              mvp[0], mvp[1], 8, 8 );
         }
 

diff --git a/encoder/me.c b/encoder/me.c
index 374cf8124c326d5d7d444875b4188ff91652c0fb..9adcc6cbd714b8853aa9dfff0089f16c94360e0e 100644 (file)
--- a/encoder/me.c
+++ b/encoder/me.c
@@ -63,7 +63,7 @@ static void refine_subpel( x264_t *h, x264_me_t *m, int hpel_iters, int qpel_ite
 #define COST_MV_HPEL( mx, my ) \
 { \
     int stride = 16; \
-    uint8_t *src = h->mc.get_ref( m->p_fref, m->i_stride[0], pix, &stride, mx, my, bw, bh ); \
+    uint8_t *src = h->mc.get_ref( pix, &stride, m->p_fref, m->i_stride[0], mx, my, bw, bh ); \
     int cost = h->pixf.sad[i_pixel]( m->p_fenc[0], FENC_STRIDE, src, stride ) \
              + p_cost_mvx[ mx ] + p_cost_mvy[ my ]; \
     COPY3_IF_LT( bpred_cost, cost, bpred_mx, mx, bpred_my, my ); \
@@ -552,7 +552,7 @@ void x264_me_refine_qpel( x264_t *h, x264_me_t *m )
 #define COST_MV_SAD( mx, my ) \
 { \
     int stride = 16; \
-    uint8_t *src = h->mc.get_ref( m->p_fref, m->i_stride[0], pix[0], &stride, mx, my, bw, bh ); \
+    uint8_t *src = h->mc.get_ref( pix[0], &stride, m->p_fref, m->i_stride[0], mx, my, bw, bh ); \
     int cost = h->pixf.sad[i_pixel]( m->p_fenc[0], FENC_STRIDE, src, stride ) \
              + p_cost_mvx[ mx ] + p_cost_mvy[ my ]; \
     COPY3_IF_LT( bcost, cost, bmx, mx, bmy, my ); \
@@ -562,16 +562,16 @@ void x264_me_refine_qpel( x264_t *h, x264_me_t *m )
 if( b_refine_qpel || (dir^1) != odir ) \
 { \
     int stride = 16; \
-    uint8_t *src = h->mc.get_ref( m->p_fref, m->i_stride[0], pix[0], &stride, mx, my, bw, bh ); \
+    uint8_t *src = h->mc.get_ref( pix[0], &stride, m->p_fref, m->i_stride[0], mx, my, bw, bh ); \
     int cost = h->pixf.mbcmp[i_pixel]( m->p_fenc[0], FENC_STRIDE, src, stride ) \
              + p_cost_mvx[ mx ] + p_cost_mvy[ my ]; \
     if( b_chroma_me && cost < bcost ) \
     { \
-        h->mc.mc_chroma( m->p_fref[4], m->i_stride[1], pix[0], 8, mx, my, bw/2, bh/2 ); \
+        h->mc.mc_chroma( pix[0], 8, m->p_fref[4], m->i_stride[1], mx, my, bw/2, bh/2 ); \
         cost += h->pixf.mbcmp[i_pixel+3]( m->p_fenc[1], FENC_STRIDE, pix[0], 8 ); \
         if( cost < bcost ) \
         { \
-            h->mc.mc_chroma( m->p_fref[5], m->i_stride[1], pix[0], 8, mx, my, bw/2, bh/2 ); \
+            h->mc.mc_chroma( pix[0], 8, m->p_fref[5], m->i_stride[1], mx, my, bw/2, bh/2 ); \
             cost += h->pixf.mbcmp[i_pixel+3]( m->p_fenc[2], FENC_STRIDE, pix[0], 8 ); \
         } \
     } \
@@ -619,8 +619,8 @@ static void refine_subpel( x264_t *h, x264_me_t *m, int hpel_iters, int qpel_ite
         int costs[4];
         int stride = 32; // candidates are either all hpel or all qpel, so one stride is enough
         uint8_t *src0, *src1, *src2, *src3;
-        src0 = h->mc.get_ref( m->p_fref, m->i_stride[0], pix[0], &stride, omx, omy-2, bw, bh+1 );
-        src2 = h->mc.get_ref( m->p_fref, m->i_stride[0], pix[1], &stride, omx-2, omy, bw+4, bh );
+        src0 = h->mc.get_ref( pix[0], &stride, m->p_fref, m->i_stride[0], omx, omy-2, bw, bh+1 );
+        src2 = h->mc.get_ref( pix[1], &stride, m->p_fref, m->i_stride[0], omx-2, omy, bw+4, bh );
         src1 = src0 + stride;
         src3 = src2 + 1;
         h->pixf.sad_x4[i_pixel]( m->p_fenc[0], src0, src1, src2, src3, stride, costs );
@@ -688,8 +688,8 @@ static void refine_subpel( x264_t *h, x264_me_t *m, int hpel_iters, int qpel_ite
 #define BIME_CACHE( dx, dy ) \
 { \
     int i = 4 + 3*dx + dy; \
-    h->mc.mc_luma( m0->p_fref, m0->i_stride[0], pix0[i], bw, om0x+dx, om0y+dy, bw, bh ); \
-    h->mc.mc_luma( m1->p_fref, m1->i_stride[0], pix1[i], bw, om1x+dx, om1y+dy, bw, bh ); \
+    h->mc.mc_luma( pix0[i], bw, m0->p_fref, m0->i_stride[0], om0x+dx, om0y+dy, bw, bh ); \
+    h->mc.mc_luma( pix1[i], bw, m1->p_fref, m1->i_stride[0], om1x+dx, om1y+dy, bw, bh ); \
 }
 
 #define BIME_CACHE2(a,b) \
@@ -803,7 +803,7 @@ int x264_me_refine_bidir( x264_t *h, x264_me_t *m0, x264_me_t *m1, int i_weight
 #define COST_MV_SATD( mx, my, dst ) \
 { \
     int stride = 16; \
-    uint8_t *src = h->mc.get_ref( m->p_fref, m->i_stride[0], pix, &stride, mx, my, bw*4, bh*4 ); \
+    uint8_t *src = h->mc.get_ref( pix, &stride, m->p_fref, m->i_stride[0], mx, my, bw*4, bh*4 ); \
     dst = h->pixf.mbcmp[i_pixel]( m->p_fenc[0], FENC_STRIDE, src, stride ) \
         + p_cost_mvx[mx] + p_cost_mvy[my]; \
     COPY1_IF_LT( bsatd, dst ); \

diff --git a/encoder/slicetype.c b/encoder/slicetype.c
index bfe44e579b678165204232a7251805c718d0e0ea..f3faa5da625936ea3dab2a3f81959a7b47858266 100644 (file)
--- a/encoder/slicetype.c
+++ b/encoder/slicetype.c
@@ -107,9 +107,9 @@ int x264_slicetype_mb_cost( x264_t *h, x264_mb_analysis_t *a,
         int stride2 = 16; \
         uint8_t *src2; \
         int i_cost; \
-        h->mc.mc_luma( m[0].p_fref, m[0].i_stride[0], pix1, 16, \
+        h->mc.mc_luma( pix1, 16, m[0].p_fref, m[0].i_stride[0], \
                        (mv0)[0], (mv0)[1], 8, 8 ); \
-        src2 = h->mc.get_ref( m[1].p_fref, m[1].i_stride[0], pix2, &stride2, \
+        src2 = h->mc.get_ref( pix2, &stride2, m[1].p_fref, m[1].i_stride[0], \
                        (mv1)[0], (mv1)[1], 8, 8 ); \
         h->mc.avg[PIXEL_8x8]( pix1, 16, src2, stride2 ); \
         i_cost = penalty + h->pixf.mbcmp[PIXEL_8x8]( \

diff --git a/tools/checkasm.c b/tools/checkasm.c
index 07c9210f39077ec2e9869a1303d09ac8d3826d3c..beddffc213a25cc83b313e787e06b66469d68bad 100644 (file)
--- a/tools/checkasm.c
+++ b/tools/checkasm.c
@@ -301,6 +301,7 @@ static int check_mc( int cpu_ref, int cpu_new )
     x264_mc_functions_t mc_c;
     x264_mc_functions_t mc_ref;
     x264_mc_functions_t mc_a;
+    x264_pixel_function_t pixel;
 
     uint8_t *src     = &buf1[2*32+2];
     uint8_t *src2[4] = { &buf1[2*32+2],  &buf1[7*32+2],
@@ -314,6 +315,7 @@ static int check_mc( int cpu_ref, int cpu_new )
     x264_mc_init( 0, &mc_c );
     x264_mc_init( cpu_ref, &mc_ref );
     x264_mc_init( cpu_new, &mc_a );
+    x264_pixel_init( 0, &pixel );
 
 #define MC_TEST_LUMA( w, h ) \
         if( mc_a.mc_luma != mc_ref.mc_luma ) \
@@ -321,11 +323,26 @@ static int check_mc( int cpu_ref, int cpu_new )
             used_asm = 1; \
             memset(buf3, 0xCD, 1024); \
             memset(buf4, 0xCD, 1024); \
-            mc_c.mc_luma( src2, 32, dst1, 16, dx, dy, w, h );     \
-            mc_a.mc_luma( src2, 32, dst2, 16, dx, dy, w, h );   \
-            if( memcmp( buf3, buf4, 1024 ) )               \
+            mc_c.mc_luma( dst1, 16, src2, 32, dx, dy, w, h ); \
+            mc_a.mc_luma( dst2, 16, src2, 32, dx, dy, w, h ); \
+            if( memcmp( buf3, buf4, 1024 ) ) \
             { \
-                fprintf( stderr, "mc_luma[mv(%d,%d) %2dx%-2d]     [FAILED]\n", dx, dy, w, h );   \
+                fprintf( stderr, "mc_luma[mv(%d,%d) %2dx%-2d]     [FAILED]\n", dx, dy, w, h ); \
+                ok = 0; \
+            } \
+        } \
+        if( mc_a.get_ref != mc_ref.get_ref ) \
+        { \
+            uint8_t *ref = dst2; \
+            int ref_stride = 16; \
+            used_asm = 1; \
+            memset(buf3, 0xCD, 1024); \
+            memset(buf4, 0xCD, 1024); \
+            mc_c.mc_luma( dst1, 16, src2, 32, dx, dy, w, h ); \
+            ref = mc_a.get_ref( ref, &ref_stride, src2, 32, dx, dy, w, h ); \
+            if( pixel.sad[PIXEL_##w##x##h]( dst1, 16, ref, ref_stride ) ) \
+            { \
+                fprintf( stderr, "get_ref[mv(%d,%d) %2dx%-2d]     [FAILED]\n", dx, dy, w, h ); \
                 ok = 0; \
             } \
         }
@@ -336,21 +353,21 @@ static int check_mc( int cpu_ref, int cpu_new )
             used_asm = 1; \
             memset(buf3, 0xCD, 1024); \
             memset(buf4, 0xCD, 1024); \
-            mc_c.mc_chroma( src, 32, dst1, 16, dx, dy, w, h );     \
-            mc_a.mc_chroma( src, 32, dst2, 16, dx, dy, w, h );   \
+            mc_c.mc_chroma( dst1, 16, src, 32, dx, dy, w, h ); \
+            mc_a.mc_chroma( dst2, 16, src, 32, dx, dy, w, h ); \
             /* mc_chroma width=2 may write garbage to the right of dst. ignore that. */\
             for( j=0; j<h; j++ ) \
                 for( i=w; i<4; i++ ) \
                     dst2[i+j*16] = dst1[i+j*16]; \
-            if( memcmp( buf3, buf4, 1024 ) )               \
+            if( memcmp( buf3, buf4, 1024 ) ) \
             { \
-                fprintf( stderr, "mc_chroma[mv(%d,%d) %2dx%-2d]     [FAILED]\n", dx, dy, w, h );   \
+                fprintf( stderr, "mc_chroma[mv(%d,%d) %2dx%-2d]     [FAILED]\n", dx, dy, w, h ); \
                 ok = 0; \
             } \
         }
     ok = 1; used_asm = 0;
-    for( dy = 0; dy < 4; dy++ )
-        for( dx = 0; dx < 4; dx++ )
+    for( dy = -8; dy < 8; dy++ )
+        for( dx = -8; dx < 8; dx++ )
         {
             MC_TEST_LUMA( 16, 16 );
             MC_TEST_LUMA( 16, 8 );
@@ -397,6 +414,7 @@ static int check_mc( int cpu_ref, int cpu_new )
     }
     MC_TEST_AVG( avg );
     report( "mc avg :" );
+    ok = 1; used_asm = 0;
     for( w = -64; w <= 128 && ok; w++ )
         MC_TEST_AVG( avg_weight, w );
     report( "mc wpredb :" );