[x264-devel] Minor asm changes

[Jason Garrett-Glaser]

x264 | branch: master | Jason Garrett-Glaser <jason at x264.com> | Tue Feb 14 15:07:10 2012 -0800| [ac31c59a98c6c690894670b9c9af2612f799d85b] | committer: Jason Garrett-Glaser

Minor asm changes

> http://git.videolan.org/gitweb.cgi/x264.git/?a=commit;h=ac31c59a98c6c690894670b9c9af2612f799d85b
---

 common/x86/deblock-a.asm |   11 +++++------
 common/x86/quant-a.asm   |   16 ++++++++--------
 2 files changed, 13 insertions(+), 14 deletions(-)

diff --git a/common/x86/deblock-a.asm b/common/x86/deblock-a.asm
index c452e86..d03a9de 100644
--- a/common/x86/deblock-a.asm
+++ b/common/x86/deblock-a.asm
@@ -2218,10 +2218,9 @@ cglobal deblock_h_chroma_intra_mbaff, 4,6,8
 %endmacro
 
 %macro LOAD_BYTES_XMM 1
-    movu      m0, [%1-4] ; FIXME could be aligned if we changed nnz's allocation
+    movu      m2, [%1-4] ; FIXME could be aligned if we changed nnz's allocation
     movu      m1, [%1+12]
-    mova      m2, m0
-    pslldq    m0, 1
+    pslldq    m0, m2, 1
     shufps    m2, m1, q3131 ; cur nnz, all rows
     pslldq    m1, 1
     shufps    m0, m1, q3131 ; left neighbors
@@ -2278,7 +2277,7 @@ cglobal deblock_strength, 6,6
     RET
 
 %macro DEBLOCK_STRENGTH_XMM 0
-cglobal deblock_strength, 6,6,8
+cglobal deblock_strength, 6,6,7
     ; Prepare mv comparison register
     shl      r4d, 8
     add      r4d, 3 - (1<<8)
@@ -2308,9 +2307,9 @@ cglobal deblock_strength, 6,6,8
     mova      m2, [mv+4*8*2]
     mova      m1, [mv+4*8*3]
     palignr   m3, m2, [mv+4*8*2-16], 12
-    palignr   m7, m1, [mv+4*8*3-16], 12
     psubw     m2, m3
-    psubw     m1, m7
+    palignr   m3, m1, [mv+4*8*3-16], 12
+    psubw     m1, m3
     packsswb  m2, m1
 %else
     movu      m0, [mv-4+4*8*0]
diff --git a/common/x86/quant-a.asm b/common/x86/quant-a.asm
index 970811f..bbe2930 100644
--- a/common/x86/quant-a.asm
+++ b/common/x86/quant-a.asm
@@ -1311,9 +1311,9 @@ cglobal coeff_last64, 1,4
     shl r0d, 16
     or  r1d, r2d
     or  r3d, r0d
-    shl r3,  32
-    or  r1,  r3
-    not r1
+    shl  r3, 32
+    or   r1, r3
+    not  r1
     BSR rax, r1, 0x3f
     RET
 %endif
@@ -1348,14 +1348,14 @@ cglobal coeff_level_run%1,0,7
     pxor    m2, m2
     LAST_MASK %1, t5d, t0-(%1&1)*SIZEOF_DCTCOEF, t4d
 %if %1==15
-    shr   t5d, 1
+    shr    t5d, 1
 %elif %1==8
-    and   t5d, 0xff
+    and    t5d, 0xff
 %elif %1==4
-    and   t5d, 0xf
+    and    t5d, 0xf
 %endif
-    xor   t5d, (1<<%1)-1
-    mov   [t1+4], t5d
+    xor    t5d, (1<<%1)-1
+    mov [t1+4], t5d
     shl    t5d, 32-%1
     mov    t4d, %1-1
     LZCOUNT t3d, t5d, 0x1f