Update libvpx snapshot to v0.9.7-p1 (Cayuga).

source/libvpx/vp8/encoder/x86/sad_sse2.asm

Index: source/libvpx/vp8/encoder/x86/sad_sse2.asm
===================================================================
--- source/libvpx/vp8/encoder/x86/sad_sse2.asm	(revision 96967)
+++ source/libvpx/vp8/encoder/x86/sad_sse2.asm	(working copy)
@@ -21,6 +21,7 @@
     push        rbp
     mov         rbp, rsp
     SHADOW_ARGS_TO_STACK 4
+    SAVE_XMM 6
     push        rsi
     push        rdi
     ; end prolog
@@ -34,7 +35,7 @@
         lea             rcx,        [rsi+rax*8]
 
         lea             rcx,        [rcx+rax*8]
-        pxor            xmm7,       xmm7
+        pxor            xmm6,       xmm6
 
 x16x16sad_wmt_loop:
 
@@ -52,32 +53,33 @@
         punpcklbw       xmm1,       xmm3
 
         psadbw          xmm0,       xmm1
-        movq            xmm6,       QWORD PTR [rsi+rax+8]
+        movq            xmm2,       QWORD PTR [rsi+rax+8]
 
         movq            xmm3,       QWORD PTR [rdi+rdx+8]
         lea             rsi,        [rsi+rax*2]
 
         lea             rdi,        [rdi+rdx*2]
-        punpcklbw       xmm4,       xmm6
+        punpcklbw       xmm4,       xmm2
 
         punpcklbw       xmm5,       xmm3
         psadbw          xmm4,       xmm5
 
-        paddw           xmm7,       xmm0
-        paddw           xmm7,       xmm4
+        paddw           xmm6,       xmm0
+        paddw           xmm6,       xmm4
 
         cmp             rsi,        rcx
         jne             x16x16sad_wmt_loop
 
-        movq            xmm0,       xmm7
-        psrldq          xmm7,       8
+        movq            xmm0,       xmm6
+        psrldq          xmm6,       8
 
-        paddw           xmm0,       xmm7
+        paddw           xmm0,       xmm6
         movq            rax,        xmm0
 
     ; begin epilog
     pop rdi
     pop rsi
+    RESTORE_XMM
     UNSHADOW_ARGS
     pop         rbp
     ret
@@ -112,7 +114,7 @@
 x8x16sad_wmt_loop:
 
         movq            rax,        mm7
-        cmp             rax,        arg(4)
+        cmp             eax,        arg(4)
         jg              x8x16sad_wmt_early_exit
 
         movq            mm0,        QWORD PTR [rsi]
@@ -173,7 +175,7 @@
 x8x8sad_wmt_loop:
 
         movq            rax,        mm7
-        cmp             rax,        arg(4)
+        cmp             eax,        arg(4)
         jg              x8x8sad_wmt_early_exit
 
         movq            mm0,        QWORD PTR [rsi]
@@ -282,7 +284,7 @@
 x16x8sad_wmt_loop:
 
         movq            rax,        mm7
-        cmp             rax,        arg(4)
+        cmp             eax,        arg(4)
         jg              x16x8sad_wmt_early_exit
 
         movq            mm0,        QWORD PTR [rsi]
@@ -326,3 +328,83 @@
     UNSHADOW_ARGS
     pop         rbp
     ret
+
+;void vp8_copy32xn_sse2(
+;    unsigned char *src_ptr,
+;    int  src_stride,
+;    unsigned char *dst_ptr,
+;    int  dst_stride,
+;    int height);
+global sym(vp8_copy32xn_sse2)
+sym(vp8_copy32xn_sse2):
+    push        rbp
+    mov         rbp, rsp
+    SHADOW_ARGS_TO_STACK 5
+    SAVE_XMM 7
+    push        rsi
+    push        rdi
+    ; end prolog
+
+        mov             rsi,        arg(0) ;src_ptr
+        mov             rdi,        arg(2) ;dst_ptr
+
+        movsxd          rax,        dword ptr arg(1) ;src_stride
+        movsxd          rdx,        dword ptr arg(3) ;dst_stride
+        movsxd          rcx,        dword ptr arg(4) ;height
+
+block_copy_sse2_loopx4:
+        movdqu          xmm0,       XMMWORD PTR [rsi]
+        movdqu          xmm1,       XMMWORD PTR [rsi + 16]
+        movdqu          xmm2,       XMMWORD PTR [rsi + rax]
+        movdqu          xmm3,       XMMWORD PTR [rsi + rax + 16]
+
+        lea             rsi,        [rsi+rax*2]
+
+        movdqu          xmm4,       XMMWORD PTR [rsi]
+        movdqu          xmm5,       XMMWORD PTR [rsi + 16]
+        movdqu          xmm6,       XMMWORD PTR [rsi + rax]
+        movdqu          xmm7,       XMMWORD PTR [rsi + rax + 16]
+
+        lea             rsi,    [rsi+rax*2]
+
+        movdqa          XMMWORD PTR [rdi], xmm0
+        movdqa          XMMWORD PTR [rdi + 16], xmm1
+        movdqa          XMMWORD PTR [rdi + rdx], xmm2
+        movdqa          XMMWORD PTR [rdi + rdx + 16], xmm3
+
+        lea             rdi,    [rdi+rdx*2]
+
+        movdqa          XMMWORD PTR [rdi], xmm4
+        movdqa          XMMWORD PTR [rdi + 16], xmm5
+        movdqa          XMMWORD PTR [rdi + rdx], xmm6
+        movdqa          XMMWORD PTR [rdi + rdx + 16], xmm7
+
+        lea             rdi,    [rdi+rdx*2]
+
+        sub             rcx,     4
+        cmp             rcx,     4
+        jge             block_copy_sse2_loopx4
+
+        cmp             rcx, 0
+        je              copy_is_done
+
+block_copy_sse2_loop:
+        movdqu          xmm0,       XMMWORD PTR [rsi]
+        movdqu          xmm1,       XMMWORD PTR [rsi + 16]
+        lea             rsi,    [rsi+rax]
+
+        movdqa          XMMWORD PTR [rdi], xmm0
+        movdqa          XMMWORD PTR [rdi + 16], xmm1
+        lea             rdi,    [rdi+rdx]
+
+        sub             rcx,     1
+        jne             block_copy_sse2_loop
+
+copy_is_done:
+    ; begin epilog
+    pop rdi
+    pop rsi
+    RESTORE_XMM
+    UNSHADOW_ARGS
+    pop         rbp
+    ret