separate pixel_avg into cases for mc and for bipred

git-svn-id: svn://svn.videolan.org/x264/trunk@693 df754926-b1dd-0310-bc7b-ec298dee348c

diff --git a/common/amd64/mc-a.asm b/common/amd64/mc-a.asm
index f1b4bea06824b639ce81b399248156efe86cd908..cafce5083cde44dc287247814d0377f8d4c84c97 100644 (file)
--- a/common/amd64/mc-a.asm
+++ b/common/amd64/mc-a.asm
@@ -61,135 +61,163 @@ SECTION .text
 ;=============================================================================
 
 ;-----------------------------------------------------------------------------
-; void x264_pixel_avg_w4_mmxext( uint8_t *dst,  int i_dst_stride,
-;                                uint8_t *src1, int i_src1_stride,
-;                                uint8_t *src2, int i_src2_stride,
-;                                int i_height );
+; void x264_pixel_avg_w4_mmxext( uint8_t *dst, int dst_stride,
+;                                uint8_t *src, int src_stride,
+;                                int height );
 ;-----------------------------------------------------------------------------
 cglobal x264_pixel_avg_w4_mmxext
-    mov         r10, parm5q         ; src2
-    movsxd      r11, parm6d         ; i_src2_stride
-    mov         eax, parm7d         ; i_height
-
-ALIGN 4
-.height_loop    
-    movd        mm0, [parm3q]
-    movd        mm1, [parm3q+parm4q]
-    pavgb       mm0, [r10]
-    pavgb       mm1, [r10+r11]
-    movd        [parm1q], mm0
-    movd        [parm1q+parm2q], mm1
-    sub         eax, 2
-    lea         parm3q, [parm3q+parm4q*2]
-    lea         r10, [r10+r11*2]
-    lea         parm1q, [parm1q+parm2q*2]
-    jg          .height_loop
+.height_loop:
+    movd   mm0, [parm3q]
+    movd   mm1, [parm3q+parm4q]
+    pavgb  mm0, [parm1q]
+    pavgb  mm1, [parm1q+parm2q]
+    movd   [parm1q], mm0
+    movd   [parm1q+parm2q], mm1
+    sub    temp1d, 2
+    lea    parm3q, [parm3q+parm4q*2]
+    lea    parm1q, [parm1q+parm2q*2]
+    jg     .height_loop
     rep ret
 
-                          
-
-;-----------------------------------------------------------------------------
-; void x264_pixel_avg_w8_mmxext( uint8_t *dst,  int i_dst_stride,
-;                                uint8_t *src1, int i_src1_stride,
-;                                uint8_t *src2, int i_src2_stride,
-;                                int i_height );
-;-----------------------------------------------------------------------------
 cglobal x264_pixel_avg_w8_mmxext
-    mov         r10, parm5q         ; src2
-    movsxd      r11, parm6d         ; i_src2_stride
-    mov         eax, parm7d         ; i_height
-
-ALIGN 4
-.height_loop    
-    movq        mm0, [parm3q]
-    movq        mm1, [parm3q+parm4q]
-    pavgb       mm0, [r10]
-    pavgb       mm1, [r10+r11]
-    movq        [parm1q], mm0
-    movq        [parm1q+parm2q], mm1
-    sub         eax, 2
-    lea         parm3q, [parm3q+parm4q*2]
-    lea         r10, [r10+r11*2]
-    lea         parm1q, [parm1q+parm2q*2]
-    jg          .height_loop
+.height_loop:
+    movq   mm0, [parm3q]
+    movq   mm1, [parm3q+parm4q]
+    pavgb  mm0, [parm1q]
+    pavgb  mm1, [parm1q+parm2q]
+    movq   [parm1q], mm0
+    movq   [parm1q+parm2q], mm1
+    sub    temp1d, 2
+    lea    parm3q, [parm3q+parm4q*2]
+    lea    parm1q, [parm1q+parm2q*2]
+    jg     .height_loop
     rep ret
 
-;-----------------------------------------------------------------------------
-; void x264_pixel_avg_w16_mmxext( uint8_t *dst,  int i_dst_stride,
-;                                 uint8_t *src1, int i_src1_stride,
-;                                 uint8_t *src2, int i_src2_stride,
-;                                 int i_height );
-;-----------------------------------------------------------------------------
 cglobal x264_pixel_avg_w16_mmxext
-    mov         r10, parm5q         ; src2
-    movsxd      r11, parm6d         ; i_src2_stride
-    mov         eax, parm7d         ; i_height
+.height_loop:
+    movq   mm0, [parm3q  ]
+    movq   mm1, [parm3q+8]
+    movq   mm2, [parm3q+parm4q  ]
+    movq   mm3, [parm3q+parm4q+8]
+    pavgb  mm0, [parm1q  ]
+    pavgb  mm1, [parm1q+8]
+    pavgb  mm2, [parm1q+parm2q  ]
+    pavgb  mm3, [parm1q+parm2q+8]
+    movq   [parm1q  ], mm0
+    movq   [parm1q+8], mm1
+    movq   [parm1q+parm2q  ], mm2
+    movq   [parm1q+parm2q+8], mm3
+    sub    temp1d, 2
+    lea    parm3q, [parm3q+parm4q*2]
+    lea    parm1q, [parm1q+parm2q*2]
+    jg     .height_loop
+    rep ret
 
-ALIGN 4
-.height_loop    
-    movq        mm0, [parm3q  ]
-    movq        mm1, [parm3q+8]
-    pavgb       mm0, [r10  ]
-    pavgb       mm1, [r10+8]
-    movq        [parm1q  ], mm0
-    movq        [parm1q+8], mm1
-    dec         eax
-    lea         parm3q, [parm3q+parm4q]
-    lea         r10, [r10+r11]
-    lea         parm1q, [parm1q+parm2q]
-    jg          .height_loop
+cglobal x264_pixel_avg_w16_sse2
+.height_loop:
+    movdqu xmm0, [parm3q]
+    movdqu xmm1, [parm3q+parm4q]
+    pavgb  xmm0, [parm1q]
+    pavgb  xmm1, [parm1q+parm2q]
+    movdqa [parm1q], xmm0
+    movdqa [parm1q+parm2q], xmm1
+    sub    temp1d, 2
+    lea    parm3q, [parm3q+parm4q*2]
+    lea    parm1q, [parm1q+parm2q*2]
+    jg     .height_loop
     rep ret
 
+%macro AVGH 2
+cglobal x264_pixel_avg_%1x%2_mmxext
+    mov temp1d, %2
+    jmp x264_pixel_avg_w%1_mmxext
+%endmacro
+
+AVGH 16, 16
+AVGH 16, 8
+AVGH 8, 16
+AVGH 8, 8
+AVGH 8, 4
+AVGH 4, 8
+AVGH 4, 4
+AVGH 4, 2
+
 ;-----------------------------------------------------------------------------
-; void x264_pixel_avg_w20_mmxext( uint8_t *dst,  int i_dst_stride,
-;                                 uint8_t *src1, int i_src1_stride,
-;                                 uint8_t *src2, int i_src2_stride,
-;                                 int i_height );
+; void x264_pixel_avg2_w4_mmxext( uint8_t *dst, int dst_stride,
+;                                 uint8_t *src1, int src_stride,
+;                                 uint8_t *src2, int height );
 ;-----------------------------------------------------------------------------
-cglobal x264_pixel_avg_w20_mmxext
-    mov         r10, parm5q         ; src2
-    movsxd      r11, parm6d         ; i_src2_stride
-    mov         eax, parm7d         ; i_height
+%macro AVG2_START 0
+%ifdef WIN64
+    mov    temp1d, parm6d
+    mov    temp2q, parm5q
+%endif
+    sub    parm5q, parm3q
+%endmacro
 
-ALIGN 4
-.height_loop    
-    movq        mm0, [parm3q   ]
-    movq        mm1, [parm3q+8 ]
-    movd        mm2, [parm3q+16]
-    pavgb       mm0, [r10   ]
-    pavgb       mm1, [r10+8 ]
-    pavgb       mm2, [r10+16]
-    movq        [parm1q   ], mm0
-    movq        [parm1q+8 ], mm1
-    movd        [parm1q+16], mm2
-    dec         eax
-    lea         parm3q, [parm3q+parm4q]
-    lea         r10, [r10+r11]
-    lea         parm1q, [parm1q+parm2q]
-    jg          .height_loop
+cglobal x264_pixel_avg2_w4_mmxext
+    AVG2_START
+    lea    r10, [temp2q+parm4q]
+.height_loop:
+    movd   mm0, [parm3q]
+    movd   mm1, [parm3q+parm4q]
+    pavgb  mm0, [parm3q+temp2q]
+    pavgb  mm1, [parm3q+r10]
+    movd   [parm1q], mm0
+    movd   [parm1q+parm2q], mm1
+    sub    temp1d, 2
+    lea    parm3q, [parm3q+parm4q*2]
+    lea    parm1q, [parm1q+parm2q*2]
+    jg     .height_loop
     rep ret
 
-;-----------------------------------------------------------------------------
-; void x264_pixel_avg_w16_sse2( uint8_t *dst,  int i_dst_stride,
-;                               uint8_t *src1, int i_src1_stride,
-;                               uint8_t *src2, int i_src2_stride,
-;                               int i_height );
-;-----------------------------------------------------------------------------
-cglobal x264_pixel_avg_w16_sse2
-    mov         r10, parm5q         ; src2
-    movsxd      r11, parm6d         ; i_src2_stride
-    mov         eax, parm7d         ; i_height
+cglobal x264_pixel_avg2_w8_mmxext
+    AVG2_START
+    lea    r10, [temp2q+parm4q]
+.height_loop:
+    movq   mm0, [parm3q]
+    movq   mm1, [parm3q+parm4q]
+    pavgb  mm0, [parm3q+temp2q]
+    pavgb  mm1, [parm3q+r10]
+    movq   [parm1q], mm0
+    movq   [parm1q+parm2q], mm1
+    sub    temp1d, 2
+    lea    parm3q, [parm3q+parm4q*2]
+    lea    parm1q, [parm1q+parm2q*2]
+    jg     .height_loop
+    rep ret
 
-ALIGN 4
-.height_loop    
-    movdqu      xmm0, [parm3q]
-    pavgb       xmm0, [r10]
-    movdqu      [parm1q], xmm0
-    dec         rax
-    lea         parm3q, [parm3q+parm4q]
-    lea         r10, [r10+r11]
-    lea         parm1q, [parm1q+parm2q]
-    jg          .height_loop
+cglobal x264_pixel_avg2_w16_mmxext
+    AVG2_START
+.height_loop:
+    movq   mm0, [parm3q]
+    movq   mm1, [parm3q+8]
+    pavgb  mm0, [parm3q+temp2q]
+    pavgb  mm1, [parm3q+temp2q+8]
+    movq   [parm1q], mm0
+    movq   [parm1q+8], mm1
+    add    parm3q, parm4q
+    add    parm1q, parm2q
+    dec    temp1d
+    jg     .height_loop
+    rep ret
+
+cglobal x264_pixel_avg2_w20_mmxext
+    AVG2_START
+.height_loop:
+    movq   mm0, [parm3q]
+    movq   mm1, [parm3q+8]
+    movd   mm2, [parm3q+16]
+    pavgb  mm0, [parm3q+temp2q]
+    pavgb  mm1, [parm3q+temp2q+8]
+    pavgb  mm2, [parm3q+temp2q+16]
+    movq   [parm1q], mm0
+    movq   [parm1q+8], mm1
+    movd   [parm1q+16], mm2
+    add    parm3q, parm4q
+    add    parm1q, parm2q
+    dec    temp1d
+    jg     .height_loop
     rep ret
 
 

diff --git a/common/i386/mc-a.asm b/common/i386/mc-a.asm
index c689a8df722c2cb9f5513bfab478286895e65e7c..719107391d0a7a5425aeadd4f7c87041f782d17b 100644 (file)
--- a/common/i386/mc-a.asm
+++ b/common/i386/mc-a.asm
@@ -61,208 +61,173 @@ SECTION .text
 ;=============================================================================
 
 ;-----------------------------------------------------------------------------
-; void x264_pixel_avg_w4_mmxext( uint8_t *dst,  int i_dst_stride,
-;                                uint8_t *src1, int i_src1_stride,
-;                                uint8_t *src2, int i_src2_stride,
-;                                int i_height );
+; void x264_pixel_avg_w4_mmxext( uint8_t *dst, int dst_stride,
+;                                uint8_t *src, int src_stride,
+;                                int height );
 ;-----------------------------------------------------------------------------
-cglobal x264_pixel_avg_w4_mmxext
-    push        ebp
-    push        ebx
-    push        esi
-    push        edi
-
-    mov         edi, [esp+20]       ; dst
-    mov         ebx, [esp+28]       ; src1
-    mov         ecx, [esp+36]       ; src2
-    mov         esi, [esp+24]       ; i_dst_stride
-    mov         eax, [esp+32]       ; i_src1_stride
-    mov         edx, [esp+40]       ; i_src2_stride
-    mov         ebp, [esp+44]       ; i_height
-ALIGN 4
-.height_loop    
-    movd        mm0, [ebx]
-    pavgb       mm0, [ecx]
-    movd        mm1, [ebx+eax]
-    pavgb       mm1, [ecx+edx]
-    movd        [edi], mm0
-    movd        [edi+esi], mm1
-    dec         ebp
-    dec         ebp
-    lea         ebx, [ebx+eax*2]
-    lea         ecx, [ecx+edx*2]
-    lea         edi, [edi+esi*2]
-    jg          .height_loop
-
-    pop         edi
-    pop         esi
-    pop         ebx
-    pop         ebp
-    ret
-
-                          
-
-;-----------------------------------------------------------------------------
-; void x264_pixel_avg_w8_mmxext( uint8_t *dst,  int i_dst_stride,
-;                                uint8_t *src1, int i_src1_stride,
-;                                uint8_t *src2, int i_src2_stride,
-;                                int i_height );
-;-----------------------------------------------------------------------------
-cglobal x264_pixel_avg_w8_mmxext
-    push        ebp
-    push        ebx
-    push        esi
-    push        edi
-
-    mov         edi, [esp+20]       ; dst
-    mov         ebx, [esp+28]       ; src1
-    mov         ecx, [esp+36]       ; src2
-    mov         esi, [esp+24]       ; i_dst_stride
-    mov         eax, [esp+32]       ; i_src1_stride
-    mov         edx, [esp+40]       ; i_src2_stride
-    mov         ebp, [esp+44]       ; i_height
-ALIGN 4
-.height_loop    
-    movq        mm0, [ebx]
-    pavgb       mm0, [ecx]
-    movq        [edi], mm0
-    dec         ebp
-    lea         ebx, [ebx+eax]
-    lea         ecx, [ecx+edx]
-    lea         edi, [edi+esi]
-    jg          .height_loop
-
-    pop         edi
-    pop         esi
-    pop         ebx
-    pop         ebp
-    ret
-
-
+%macro AVG_START 1
+cglobal %1
+    push   ebx
+    mov    eax, [esp+12]  ; dst
+    mov    ebx, [esp+16]  ; dst_stride
+    mov    ecx, [esp+20]  ; src
+    mov    edx, [esp+24]  ; src_stride
+    ; esi = height
+.height_loop:
+%endmacro
 
-;-----------------------------------------------------------------------------
-; void x264_pixel_avg_w16_mmxext( uint8_t *dst,  int i_dst_stride,
-;                                 uint8_t *src1, int i_src1_stride,
-;                                 uint8_t *src2, int i_src2_stride,
-;                                 int i_height );
-;-----------------------------------------------------------------------------
-cglobal x264_pixel_avg_w16_mmxext
-    push        ebp
-    push        ebx
-    push        esi
-    push        edi
-
-    mov         edi, [esp+20]       ; dst
-    mov         ebx, [esp+28]       ; src1
-    mov         ecx, [esp+36]       ; src2
-    mov         esi, [esp+24]       ; i_dst_stride
-    mov         eax, [esp+32]       ; i_src1_stride
-    mov         edx, [esp+40]       ; i_src2_stride
-    mov         ebp, [esp+44]       ; i_height
-ALIGN 4
-.height_loop    
-    movq        mm0, [ebx  ]
-    movq        mm1, [ebx+8]
-    pavgb       mm0, [ecx  ]
-    pavgb       mm1, [ecx+8]
-    movq        [edi  ], mm0
-    movq        [edi+8], mm1
-    dec         ebp
-    lea         ebx, [ebx+eax]
-    lea         ecx, [ecx+edx]
-    lea         edi, [edi+esi]
-    jg          .height_loop
-
-    pop         edi
-    pop         esi
-    pop         ebx
-    pop         ebp
+%macro AVG_END 0
+    sub    esi, 2
+    lea    eax, [eax+ebx*2]
+    lea    ecx, [ecx+edx*2]
+    jg     .height_loop
+    pop    ebx
+    pop    esi
     ret
+%endmacro
 
+AVG_START x264_pixel_avg_w4_mmxext
+    movd   mm0, [ecx]
+    movd   mm1, [ecx+edx]
+    pavgb  mm0, [eax]
+    pavgb  mm1, [eax+ebx]
+    movd   [eax], mm0
+    movd   [eax+ebx], mm1
+AVG_END
+
+AVG_START x264_pixel_avg_w8_mmxext
+    movq   mm0, [ecx]
+    movq   mm1, [ecx+edx]
+    pavgb  mm0, [eax]
+    pavgb  mm1, [eax+ebx]
+    movq   [eax], mm0
+    movq   [eax+ebx], mm1
+AVG_END
+
+AVG_START x264_pixel_avg_w16_mmxext
+    movq   mm0, [ecx]
+    movq   mm1, [ecx+8]
+    movq   mm2, [ecx+edx]
+    movq   mm3, [ecx+edx+8]
+    pavgb  mm0, [eax]
+    pavgb  mm1, [eax+8]
+    pavgb  mm2, [eax+ebx]
+    pavgb  mm3, [eax+ebx+8]
+    movq   [eax], mm0
+    movq   [eax+8], mm1
+    movq   [eax+ebx], mm2
+    movq   [eax+ebx+8], mm3
+AVG_END
+
+AVG_START x264_pixel_avg_w16_sse2
+    movdqu xmm0, [ecx]
+    movdqu xmm1, [ecx+edx]
+    pavgb  xmm0, [eax]
+    pavgb  xmm1, [eax+ebx]
+    movdqa [eax], xmm0
+    movdqa [eax+ebx], xmm1
+AVG_END
+
+%macro AVGH 2
+cglobal x264_pixel_avg_%1x%2_mmxext
+    push esi
+    mov esi, %2
+    jmp x264_pixel_avg_w%1_mmxext
+%endmacro
 
+AVGH 16, 16
+AVGH 16, 8
+AVGH 8, 16
+AVGH 8, 8
+AVGH 8, 4
+AVGH 4, 8
+AVGH 4, 4
+AVGH 4, 2
+
+%macro AVG2_START 1
+cglobal %1
+    push   ebx
+    push   esi
+    push   edi
+    push   ebp
+    mov    eax, [esp+20]  ; dst
+    mov    ebx, [esp+24]  ; dst_stride
+    mov    ecx, [esp+28]  ; src1
+    mov    edx, [esp+32]  ; src_stride
+    mov    edi, [esp+36]  ; src2
+    mov    esi, [esp+40]  ; height
+    sub    edi, ecx
+    lea    ebp, [edi+edx]
+.height_loop:
+%endmacro
 
-;-----------------------------------------------------------------------------
-; void x264_pixel_avg_w20_mmxext( uint8_t *dst,  int i_dst_stride,
-;                                 uint8_t *src1, int i_src1_stride,
-;                                 uint8_t *src2, int i_src2_stride,
-;                                 int i_height );
-;-----------------------------------------------------------------------------
-cglobal x264_pixel_avg_w20_mmxext
-    push        ebp
-    push        ebx
-    push        esi
-    push        edi
-
-    mov         edi, [esp+20]       ; dst
-    mov         ebx, [esp+28]       ; src1
-    mov         ecx, [esp+36]       ; src2
-    mov         esi, [esp+24]       ; i_dst_stride
-    mov         eax, [esp+32]       ; i_src1_stride
-    mov         edx, [esp+40]       ; i_src2_stride
-    mov         ebp, [esp+44]       ; i_height
-ALIGN 4
-.height_loop    
-    movq        mm0, [ebx   ]
-    movq        mm1, [ebx+8 ]
-    movd        mm2, [ebx+16]
-    pavgb       mm0, [ecx   ]
-    pavgb       mm1, [ecx+8 ]
-    pavgb       mm2, [ecx+16]
-    movq        [edi   ], mm0
-    movq        [edi+8 ], mm1
-    movd        [edi+16], mm2
-    dec         ebp
-    lea         ebx, [ebx+eax]
-    lea         ecx, [ecx+edx]
-    lea         edi, [edi+esi]
-    jg          .height_loop
-
-    pop         edi
-    pop         esi
-    pop         ebx
-    pop         ebp
+%macro AVG2_END 0
+    sub    esi, 2
+    lea    eax, [eax+ebx*2]
+    lea    ecx, [ecx+edx*2]
+    jg     .height_loop
+    pop    ebp
+    pop    edi
+    pop    esi
+    pop    ebx
     ret
+%endmacro
 
+AVG2_START x264_pixel_avg2_w4_mmxext
+    movd   mm0, [ecx]
+    movd   mm1, [ecx+edx]
+    pavgb  mm0, [ecx+edi]
+    pavgb  mm1, [ecx+ebp]
+    movd   [eax], mm0
+    movd   [eax+ebx], mm1
+AVG2_END
+
+AVG2_START x264_pixel_avg2_w8_mmxext
+    movq   mm0, [ecx]
+    movq   mm1, [ecx+edx]
+    pavgb  mm0, [ecx+edi]
+    pavgb  mm1, [ecx+ebp]
+    movq   [eax], mm0
+    movq   [eax+ebx], mm1
+AVG2_END
+
+AVG2_START x264_pixel_avg2_w16_mmxext
+    movq   mm0, [ecx]
+    movq   mm1, [ecx+8]
+    movq   mm2, [ecx+edx]
+    movq   mm3, [ecx+edx+8]
+    pavgb  mm0, [ecx+edi]
+    pavgb  mm1, [ecx+edi+8]
+    pavgb  mm2, [ecx+ebp]
+    pavgb  mm3, [ecx+ebp+8]
+    movq   [eax], mm0
+    movq   [eax+8], mm1
+    movq   [eax+ebx], mm2
+    movq   [eax+ebx+8], mm3
+AVG2_END
+
+AVG2_START x264_pixel_avg2_w20_mmxext
+    movq   mm0, [ecx]
+    movq   mm1, [ecx+8]
+    movd   mm2, [ecx+16]
+    movq   mm3, [ecx+edx]
+    movq   mm4, [ecx+edx+8]
+    movd   mm5, [ecx+edx+16]
+    pavgb  mm0, [ecx+edi]
+    pavgb  mm1, [ecx+edi+8]
+    pavgb  mm2, [ecx+edi+16]
+    pavgb  mm3, [ecx+ebp]
+    pavgb  mm4, [ecx+ebp+8]
+    pavgb  mm5, [ecx+ebp+16]
+    movq   [eax], mm0
+    movq   [eax+8], mm1
+    movd   [eax+16], mm2
+    movq   [eax+ebx], mm3
+    movq   [eax+ebx+8], mm4
+    movd   [eax+ebx+16], mm5
+AVG2_END
 
 
-ALIGN 16
-;-----------------------------------------------------------------------------
-; void x264_pixel_avg_w16_sse2( uint8_t *dst,  int i_dst_stride,
-;                               uint8_t *src1, int i_src1_stride,
-;                               uint8_t *src2, int i_src2_stride,
-;                               int i_height );
-;-----------------------------------------------------------------------------
-cglobal x264_pixel_avg_w16_sse2
-    push        ebp
-    push        ebx
-    push        esi
-    push        edi
-
-    mov         edi, [esp+20]       ; dst
-    mov         ebx, [esp+28]       ; src1
-    mov         ecx, [esp+36]       ; src2
-    mov         esi, [esp+24]       ; i_dst_stride
-    mov         eax, [esp+32]       ; i_src1_stride
-    mov         edx, [esp+40]       ; i_src2_stride
-    mov         ebp, [esp+44]       ; i_height
-ALIGN 4
-.height_loop    
-    movdqu      xmm0, [ebx]
-    pavgb       xmm0, [ecx]
-    movdqu      [edi], xmm0
-
-    dec         ebp
-    lea         ebx, [ebx+eax]
-    lea         ecx, [ecx+edx]
-    lea         edi, [edi+esi]
-    jg          .height_loop
-
-    pop         edi
-    pop         esi
-    pop         ebx
-    pop         ebp
-    ret
-
 
 ;=============================================================================
 ; weighted prediction

diff --git a/common/i386/mc-c.c b/common/i386/mc-c.c
index bbe15dc62c02d1d9cac8809a382df5cddf3bf0c5..e60c15f3ceef12489bece310e554cdb36daec93b 100644 (file)
--- a/common/i386/mc-c.c
+++ b/common/i386/mc-c.c
@@ -28,11 +28,18 @@
 #include "common/common.h"
 
 /* NASM functions */
-extern void x264_pixel_avg_w4_mmxext( uint8_t *,  int, uint8_t *, int, uint8_t *, int, int );
-extern void x264_pixel_avg_w8_mmxext( uint8_t *,  int, uint8_t *, int, uint8_t *, int, int );
-extern void x264_pixel_avg_w16_mmxext( uint8_t *,  int, uint8_t *, int, uint8_t *, int, int );
-extern void x264_pixel_avg_w20_mmxext( uint8_t *,  int, uint8_t *, int, uint8_t *, int, int );
-extern void x264_pixel_avg_w16_sse2( uint8_t *,  int, uint8_t *, int, uint8_t *, int, int );
+extern void x264_pixel_avg_16x16_mmxext( uint8_t *, int, uint8_t *, int );
+extern void x264_pixel_avg_16x8_mmxext( uint8_t *, int, uint8_t *, int );
+extern void x264_pixel_avg_8x16_mmxext( uint8_t *, int, uint8_t *, int );
+extern void x264_pixel_avg_8x8_mmxext( uint8_t *, int, uint8_t *, int );
+extern void x264_pixel_avg_8x4_mmxext( uint8_t *, int, uint8_t *, int );
+extern void x264_pixel_avg_4x8_mmxext( uint8_t *, int, uint8_t *, int );
+extern void x264_pixel_avg_4x4_mmxext( uint8_t *, int, uint8_t *, int );
+extern void x264_pixel_avg_4x2_mmxext( uint8_t *, int, uint8_t *, int );
+extern void x264_pixel_avg2_w4_mmxext( uint8_t *, int, uint8_t *, int, uint8_t *, int );
+extern void x264_pixel_avg2_w8_mmxext( uint8_t *, int, uint8_t *, int, uint8_t *, int );
+extern void x264_pixel_avg2_w16_mmxext( uint8_t *, int, uint8_t *, int, uint8_t *, int );
+extern void x264_pixel_avg2_w20_mmxext( uint8_t *, int, uint8_t *, int, uint8_t *, int );
 extern void x264_pixel_avg_weight_4x4_mmxext( uint8_t *, int, uint8_t *, int, int );
 extern void x264_pixel_avg_weight_w8_mmxext( uint8_t *, int, uint8_t *, int, int, int );
 extern void x264_pixel_avg_weight_w16_mmxext( uint8_t *, int, uint8_t *, int, int, int );
@@ -46,20 +53,6 @@ extern void x264_prefetch_ref_mmxext( uint8_t *, int, int );
 extern void x264_hpel_filter_mmxext( uint8_t *dsth, uint8_t *dstv, uint8_t *dstc, uint8_t *src,
                                      int i_stride, int i_width, int i_height );
 
-#define AVG(W,H) \
-static void x264_pixel_avg_ ## W ## x ## H ## _mmxext( uint8_t *dst, int i_dst, uint8_t *src, int i_src ) \
-{ \
-    x264_pixel_avg_w ## W ## _mmxext( dst, i_dst, dst, i_dst, src, i_src, H ); \
-}
-AVG(16,16)
-AVG(16,8)
-AVG(8,16)
-AVG(8,8)
-AVG(8,4)
-AVG(4,8)
-AVG(4,4)
-AVG(4,2)
-
 #define AVG_WEIGHT(W,H) \
 void x264_pixel_avg_weight_ ## W ## x ## H ## _mmxext( uint8_t *dst, int i_dst, uint8_t *src, int i_src, int i_weight_dst ) \
 { \
@@ -71,14 +64,14 @@ AVG_WEIGHT(8,16)
 AVG_WEIGHT(8,8)
 AVG_WEIGHT(8,4)
 
-static void (* const x264_pixel_avg_wtab_mmxext[6])( uint8_t *, int, uint8_t *, int, uint8_t *, int, int ) =
+static void (* const x264_pixel_avg_wtab_mmxext[6])( uint8_t *, int, uint8_t *, int, uint8_t *, int ) =
 {
     NULL,
-    x264_pixel_avg_w4_mmxext,
-    x264_pixel_avg_w8_mmxext,
-    x264_pixel_avg_w16_mmxext,
-    x264_pixel_avg_w16_mmxext,
-    x264_pixel_avg_w20_mmxext,
+    x264_pixel_avg2_w4_mmxext,
+    x264_pixel_avg2_w8_mmxext,
+    x264_pixel_avg2_w16_mmxext,
+    x264_pixel_avg2_w16_mmxext,
+    x264_pixel_avg2_w20_mmxext,
 };
 static void (* const x264_mc_copy_wtab_mmx[5])( uint8_t *, int, uint8_t *, int, int ) =
 {
@@ -105,7 +98,7 @@ void mc_luma_mmx( uint8_t *src[4], int i_src_stride,
         uint8_t *src2 = src[hpel_ref1[qpel_idx]] + offset + ((mvx&3) == 3);
         x264_pixel_avg_wtab_mmxext[i_width>>2](
                 dst, i_dst_stride, src1, i_src_stride,
-                src2, i_src_stride, i_height );
+                src2, i_height );
     }
     else
     {
@@ -128,7 +121,7 @@ uint8_t *get_ref_mmx( uint8_t *src[4], int i_src_stride,
         uint8_t *src2 = src[hpel_ref1[qpel_idx]] + offset + ((mvx&3) == 3);
         x264_pixel_avg_wtab_mmxext[i_width>>2](
                 dst, *i_dst_stride, src1, i_src_stride,
-                src2, i_src_stride, i_height );
+                src2, i_height );
         return dst;
     }
     else
@@ -141,8 +134,8 @@ uint8_t *get_ref_mmx( uint8_t *src[4], int i_src_stride,
 
 void x264_mc_mmxext_init( x264_mc_functions_t *pf )
 {
-    pf->mc_luma   = mc_luma_mmx;
-    pf->get_ref   = get_ref_mmx;
+    pf->mc_luma = mc_luma_mmx;
+    pf->get_ref = get_ref_mmx;
 
     pf->avg[PIXEL_16x16] = x264_pixel_avg_16x16_mmxext;
     pf->avg[PIXEL_16x8]  = x264_pixel_avg_16x8_mmxext;
@@ -174,6 +167,4 @@ void x264_mc_mmxext_init( x264_mc_functions_t *pf )
 void x264_mc_sse2_init( x264_mc_functions_t *pf )
 {
     /* todo: use sse2 */
-    pf->mc_luma   = mc_luma_mmx;
-    pf->get_ref   = get_ref_mmx;
 }