H.264: tweak some other x86 asm for Atom

5 files changed, 48 insertions, 51 deletions

diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c
index dd318a1399..9909fdab78 100644
--- a/libavcodec/x86/dsputil_mmx.c
+++ b/libavcodec/x86/dsputil_mmx.c
@@ -456,12 +456,12 @@ static void put_pixels16_sse2(uint8_t *block, const uint8_t *pixels, int line_si
          "movdqu (%1,%3), %%xmm1        \n\t"
          "movdqu (%1,%3,2), %%xmm2      \n\t"
          "movdqu (%1,%4), %%xmm3        \n\t"
+         "lea (%1,%3,4), %1             \n\t"
          "movdqa %%xmm0, (%2)           \n\t"
          "movdqa %%xmm1, (%2,%3)        \n\t"
          "movdqa %%xmm2, (%2,%3,2)      \n\t"
          "movdqa %%xmm3, (%2,%4)        \n\t"
          "subl $4, %0                   \n\t"
-         "lea (%1,%3,4), %1             \n\t"
          "lea (%2,%3,4), %2             \n\t"
          "jnz 1b                        \n\t"
          : "+g"(h), "+r" (pixels),  "+r" (block)
@@ -478,6 +478,7 @@ static void avg_pixels16_sse2(uint8_t *block, const uint8_t *pixels, int line_si
          "movdqu (%1,%3), %%xmm1        \n\t"
          "movdqu (%1,%3,2), %%xmm2      \n\t"
          "movdqu (%1,%4), %%xmm3        \n\t"
+         "lea (%1,%3,4), %1             \n\t"
          "pavgb  (%2), %%xmm0           \n\t"
          "pavgb  (%2,%3), %%xmm1        \n\t"
          "pavgb  (%2,%3,2), %%xmm2      \n\t"
@@ -487,7 +488,6 @@ static void avg_pixels16_sse2(uint8_t *block, const uint8_t *pixels, int line_si
          "movdqa %%xmm2, (%2,%3,2)      \n\t"
          "movdqa %%xmm3, (%2,%4)        \n\t"
          "subl $4, %0                   \n\t"
-         "lea (%1,%3,4), %1             \n\t"
          "lea (%2,%3,4), %2             \n\t"
          "jnz 1b                        \n\t"
          : "+g"(h), "+r" (pixels),  "+r" (block)
diff --git a/libavcodec/x86/h264_chromamc.asm b/libavcodec/x86/h264_chromamc.asm
index 0a37994fb9..16cf2ec43e 100644
--- a/libavcodec/x86/h264_chromamc.asm
+++ b/libavcodec/x86/h264_chromamc.asm
@@ -72,17 +72,17 @@ SECTION .text
 .next4rows
     movq         mm0, [r1   ]
     movq         mm1, [r1+r2]
+    add           r1, r4
     CHROMAMC_AVG mm0, [r0   ]
     CHROMAMC_AVG mm1, [r0+r2]
     movq     [r0   ], mm0
     movq     [r0+r2], mm1
     add           r0, r4
-    add           r1, r4
     movq         mm0, [r1   ]
     movq         mm1, [r1+r2]
+    add           r1, r4
     CHROMAMC_AVG mm0, [r0   ]
     CHROMAMC_AVG mm1, [r0+r2]
-    add           r1, r4
     movq     [r0   ], mm0
     movq     [r0+r2], mm1
     add           r0, r4
@@ -472,8 +472,8 @@ cglobal %1_%2_chroma_mc8_%3, 6, 7, 8
     mov          r6d, r4d
     shl          r4d, 8
     sub           r4, r6
-    add           r4, 8           ; x*288+8 = x<<8 | (8-x)
     mov           r6, 8
+    add           r4, 8           ; x*288+8 = x<<8 | (8-x)
     sub          r6d, r5d
     imul          r6, r4          ; (8-y)*(x*255+8) = (8-y)*x<<8 | (8-y)*(8-x)
     imul         r4d, r5d         ;    y *(x*255+8) =    y *x<<8 |    y *(8-x)
@@ -481,24 +481,23 @@ cglobal %1_%2_chroma_mc8_%3, 6, 7, 8
     movd          m7, r6d
     movd          m6, r4d
     movdqa        m5, [rnd_2d_%2]
+    movq          m0, [r1  ]
+    movq          m1, [r1+1]
     pshuflw       m7, m7, 0
     pshuflw       m6, m6, 0
+    punpcklbw     m0, m1
     movlhps       m7, m7
     movlhps       m6, m6
 
-    movq          m0, [r1     ]
-    movq          m1, [r1   +1]
-    punpcklbw     m0, m1
-    add           r1, r2
 .next2rows
-    movq          m1, [r1     ]
-    movq          m2, [r1   +1]
-    movq          m3, [r1+r2  ]
-    movq          m4, [r1+r2+1]
+    movq          m1, [r1+r2*1   ]
+    movq          m2, [r1+r2*1+1]
+    movq          m3, [r1+r2*2  ]
+    movq          m4, [r1+r2*2+1]
     lea           r1, [r1+r2*2]
     punpcklbw     m1, m2
-    punpcklbw     m3, m4
     movdqa        m2, m1
+    punpcklbw     m3, m4
     movdqa        m4, m3
     pmaddubsw     m0, m7
     pmaddubsw     m1, m6
@@ -508,8 +507,8 @@ cglobal %1_%2_chroma_mc8_%3, 6, 7, 8
     paddw         m2, m5
     paddw         m1, m0
     paddw         m3, m2
-    movdqa        m0, m4
     psrlw         m1, 6
+    movdqa        m0, m4
     psrlw         m3, 6
 %ifidn %1, avg
     movq          m2, [r0   ]
@@ -576,6 +575,7 @@ cglobal %1_%2_chroma_mc8_%3, 6, 7, 8
     movq          m1, [r1+r2  ]
     movdqa        m2, m1
     movq          m3, [r1+r2*2]
+    lea           r1, [r1+r2*2]
     punpcklbw     m0, m1
     punpcklbw     m2, m3
     pmaddubsw     m0, m7
@@ -594,7 +594,6 @@ cglobal %1_%2_chroma_mc8_%3, 6, 7, 8
     movhps   [r0+r2], m0
     sub          r3d, 2
     lea           r0, [r0+r2*2]
-    lea           r1, [r1+r2*2]
     jg .next2yrows
     REP_RET
 %endmacro
@@ -607,8 +606,8 @@ cglobal %1_%2_chroma_mc4_%3, 6, 7, 0
     mov           r6, r4
     shl          r4d, 8
     sub          r4d, r6d
-    add          r4d, 8           ; x*288+8
     mov           r6, 8
+    add          r4d, 8           ; x*288+8
     sub          r6d, r5d
     imul         r6d, r4d         ; (8-y)*(x*255+8) = (8-y)*x<<8 | (8-y)*(8-x)
     imul         r4d, r5d         ;    y *(x*255+8) =    y *x<<8 |    y *(8-x)
@@ -616,17 +615,16 @@ cglobal %1_%2_chroma_mc4_%3, 6, 7, 0
     movd          m7, r6d
     movd          m6, r4d
     movq          m5, [pw_32]
+    movd          m0, [r1  ]
     pshufw        m7, m7, 0
+    punpcklbw     m0, [r1+1]
     pshufw        m6, m6, 0
 
-    movd          m0, [r1     ]
-    punpcklbw     m0, [r1   +1]
-    add           r1, r2
 .next2rows
-    movd          m1, [r1     ]
-    movd          m3, [r1+r2  ]
-    punpcklbw     m1, [r1   +1]
-    punpcklbw     m3, [r1+r2+1]
+    movd          m1, [r1+r2*1  ]
+    movd          m3, [r1+r2*2  ]
+    punpcklbw     m1, [r1+r2*1+1]
+    punpcklbw     m3, [r1+r2*2+1]
     lea           r1, [r1+r2*2]
     movq          m2, m1
     movq          m4, m3
@@ -638,8 +636,8 @@ cglobal %1_%2_chroma_mc4_%3, 6, 7, 0
     paddw         m2, m5
     paddw         m1, m0
     paddw         m3, m2
-    movq          m0, m4
     psrlw         m1, 6
+    movq          m0, m4
     psrlw         m3, 6
     packuswb      m1, m1
     packuswb      m3, m3
diff --git a/libavcodec/x86/h264_deblock.asm b/libavcodec/x86/h264_deblock.asm
index 9831ca2cd6..92f91acade 100644
--- a/libavcodec/x86/h264_deblock.asm
+++ b/libavcodec/x86/h264_deblock.asm
@@ -240,17 +240,17 @@ cextern pb_A1
 ; out: m1=p0' m2=q0'
 ; clobbers: m0,3-6
 %macro DEBLOCK_P0_Q0 0
-    pxor    m5, m1, m2   ; p0^q0
-    pand    m5, [pb_1]   ; (p0^q0)&1
     pcmpeqb m4, m4
+    pxor    m5, m1, m2   ; p0^q0
     pxor    m3, m4
+    pand    m5, [pb_1]   ; (p0^q0)&1
     pavgb   m3, m0       ; (p1 - q1 + 256)>>1
-    pavgb   m3, [pb_3]   ; (((p1 - q1 + 256)>>1)+4)>>1 = 64+2+(p1-q1)>>2
     pxor    m4, m1
+    pavgb   m3, [pb_3]   ; (((p1 - q1 + 256)>>1)+4)>>1 = 64+2+(p1-q1)>>2
     pavgb   m4, m2       ; (q0 - p0 + 256)>>1
     pavgb   m3, m5
-    paddusb m3, m4       ; d+128+33
     mova    m6, [pb_A1]
+    paddusb m3, m4       ; d+128+33
     psubusb m6, m3
     psubusb m3, [pb_A1]
     pminub  m6, m7
@@ -411,16 +411,16 @@ cglobal deblock_%2_luma_8_%1, 5,5
     LOAD_MASK r2, r3
 
     mov     r3, r4mp
+    pcmpeqb m3, m3
     movd    m4, [r3] ; tc0
     punpcklbw m4, m4
     punpcklbw m4, m4 ; tc = 4x tc0[3], 4x tc0[2], 4x tc0[1], 4x tc0[0]
     mova   [esp+%3], m4 ; tc
-    pcmpeqb m3, m3
     pcmpgtb m4, m3
+    mova    m3, [r4] ; p2
     pand    m4, m7
     mova   [esp], m4 ; mask
 
-    mova    m3, [r4] ; p2
     DIFF_GT2 m1, m3, m5, m6, m7 ; |p2-p0| > beta-1
     pand    m6, m4
     pand    m4, [esp+%3] ; tc
@@ -430,11 +430,10 @@ cglobal deblock_%2_luma_8_%1, 5,5
 
     mova    m4, [r0+2*r1] ; q2
     DIFF_GT2 m2, m4, m5, m6, m3 ; |q2-q0| > beta-1
-    mova    m5, [esp] ; mask
-    pand    m6, m5
+    pand    m6, [esp] ; mask
     mova    m5, [esp+%3] ; tc
-    pand    m5, m6
     psubb   m7, m6
+    pand    m5, m6
     mova    m3, [r0+r1]
     LUMA_Q1 m3, m4, [r0+2*r1], [r0+r1], m5, m6
 
@@ -482,10 +481,10 @@ cglobal deblock_h_luma_8_%1, 0,5
     ; transpose 16x4 -> original space  (only the middle 4 rows were changed by the filter)
     mov    r0, r0mp
     sub    r0, 2
-    lea    r1, [r0+r4]
 
     movq   m0, [pix_tmp+0x10]
     movq   m1, [pix_tmp+0x20]
+    lea    r1, [r0+r4]
     movq   m2, [pix_tmp+0x30]
     movq   m3, [pix_tmp+0x40]
     TRANSPOSE8x4B_STORE  PASS8ROWS(r0, r1, r3, r4)
diff --git a/libavcodec/x86/h264_idct.asm b/libavcodec/x86/h264_idct.asm
index 4788da98e0..37c2c90476 100644
--- a/libavcodec/x86/h264_idct.asm
+++ b/libavcodec/x86/h264_idct.asm
@@ -82,10 +82,10 @@ cglobal h264_idct_add_8_mmx, 3, 3, 0
     RET
 
 %macro IDCT8_1D 2
-    mova         m4, m5
     mova         m0, m1
-    psraw        m4, 1
     psraw        m1, 1
+    mova         m4, m5
+    psraw        m4, 1
     paddw        m4, m5
     paddw        m1, m0
     paddw        m4, m7
@@ -95,16 +95,16 @@ cglobal h264_idct_add_8_mmx, 3, 3, 0
 
     psubw        m0, m3
     psubw        m5, m3
+    psraw        m3, 1
     paddw        m0, m7
     psubw        m5, m7
-    psraw        m3, 1
     psraw        m7, 1
     psubw        m0, m3
     psubw        m5, m7
 
-    mova         m3, m4
     mova         m7, m1
     psraw        m1, 2
+    mova         m3, m4
     psraw        m3, 2
     paddw        m3, m0
     psraw        m0, 2
@@ -113,12 +113,12 @@ cglobal h264_idct_add_8_mmx, 3, 3, 0
     psubw        m0, m4
     psubw        m7, m5
 
-    mova         m4, m2
     mova         m5, m6
-    psraw        m4, 1
     psraw        m6, 1
-    psubw        m4, m5
+    mova         m4, m2
+    psraw        m4, 1
     paddw        m6, m2
+    psubw        m4, m5
 
     mova         m2, %1
     mova         m5, %2
@@ -337,7 +337,7 @@ cglobal h264_idct8_add4_8_mmx, 5, 7, 0
     test         r6, r6
     jz .skipblock
     mov         r6d, dword [r1+r5*4]
-    lea          r6, [r0+r6]
+    add          r6, r0
     add   word [r2], 32
     IDCT8_ADD_MMX_START r2  , rsp
     IDCT8_ADD_MMX_START r2+8, rsp+64
@@ -391,7 +391,7 @@ cglobal h264_idct_add16_8_mmx2, 5, 7, 0
     REP_RET
 .no_dc
     mov         r6d, dword [r1+r5*4]
-    lea          r6, [r0+r6]
+    add          r6, r0
     IDCT4_ADD    r6, r2, r3
 .skipblock
     inc          r5
@@ -414,7 +414,7 @@ cglobal h264_idct_add16intra_8_mmx, 5, 7, 0
     test         r6, r6
     jz .skipblock
     mov         r6d, dword [r1+r5*4]
-    lea          r6, [r0+r6]
+    add          r6, r0
     IDCT4_ADD    r6, r2, r3
 .skipblock
     inc          r5
@@ -456,7 +456,7 @@ cglobal h264_idct_add16intra_8_mmx2, 5, 7, 0
 %define dst_regd r1d
 %endif
     mov    dst_regd, dword [r1+r5*4]
-    lea     dst_reg, [r0+dst_reg]
+    add     dst_reg, r0
     DC_ADD_MMX2_OP movh, dst_reg, r3, r6
 %ifndef ARCH_X86_64
     mov          r1, r1m
@@ -513,7 +513,7 @@ cglobal h264_idct8_add4_8_mmx2, 5, 7, 0
     RET
 .no_dc
     mov         r6d, dword [r1+r5*4]
-    lea          r6, [r0+r6]
+    add          r6, r0
     add   word [r2], 32
     IDCT8_ADD_MMX_START r2  , rsp
     IDCT8_ADD_MMX_START r2+8, rsp+64
@@ -558,7 +558,7 @@ INIT_MMX
 %define dst_regd r1d
 %endif
     mov    dst_regd, dword [r1+r5*4]
-    lea     dst_reg, [r0+dst_reg]
+    add     dst_reg, r0
     DC_ADD_MMX2_OP mova, dst_reg, r3, r6
     lea     dst_reg, [dst_reg+r3*4]
     DC_ADD_MMX2_OP mova, dst_reg, r3, r6
@@ -573,7 +573,7 @@ INIT_MMX
 .no_dc
 INIT_XMM
     mov    dst_regd, dword [r1+r5*4]
-    lea     dst_reg, [r0+dst_reg]
+    add     dst_reg, r0
     IDCT8_ADD_SSE dst_reg, r2, r3, r6
 %ifndef ARCH_X86_64
     mov          r1, r1m
diff --git a/libavcodec/x86/x86util.asm b/libavcodec/x86/x86util.asm
index 45196625fe..5f39107379 100644
--- a/libavcodec/x86/x86util.asm
+++ b/libavcodec/x86/x86util.asm
@@ -497,10 +497,10 @@
 %macro STORE_DIFFx2 8 ; add1, add2, reg1, reg2, zero, shift, source, stride
     movh       %3, [%7]
     movh       %4, [%7+%8]
-    punpcklbw  %3, %5
-    punpcklbw  %4, %5
     psraw      %1, %6
     psraw      %2, %6
+    punpcklbw  %3, %5
+    punpcklbw  %4, %5
     paddw      %3, %1
     paddw      %4, %2
     packuswb   %3, %5