libvpx: Pull from upstream

source/libvpx/vp8/common/x86/variance_impl_mmx.asm

 ;
 ;  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
 ;
 ;  Use of this source code is governed by a BSD-style license
 ;  that can be found in the LICENSE file in the root of the source
 ;  tree. An additional intellectual property rights grant can be found
 ;  in the file PATENTS.  All contributing project authors may
 ;  be found in the AUTHORS file in the root of the source tree.
 ;
 
@@ skipping 324 matching lines @@
         pxor        mm5, mm5                    ; Blank mmx6
         pxor        mm6, mm6                    ; Blank mmx7
         pxor        mm7, mm7                    ; Blank mmx7
 
         mov         rax, arg(0) ;[src_ptr]  ; Load base addresses
         mov         rbx, arg(2) ;[ref_ptr]
         movsxd      rcx, dword ptr arg(1) ;[source_stride]
         movsxd      rdx, dword ptr arg(3) ;[recon_stride]
 
         ; Row 1
-        movq        mm0, [rax]                  ; Copy eight bytes to mm0
-        movq        mm1, [rbx]                  ; Copy eight bytes to mm1
+        movd        mm0, [rax]                  ; Copy four bytes to mm0
+        movd        mm1, [rbx]                  ; Copy four bytes to mm1
         punpcklbw   mm0, mm6                    ; unpack to higher prrcision
         punpcklbw   mm1, mm6
         psubsw      mm0, mm1                    ; A-B (low order) to MM0
         paddw       mm5, mm0                    ; accumulate differences in mm5
         pmaddwd     mm0, mm0                    ; square and accumulate
         add         rbx,rdx                     ; Inc pointer into ref data
         add         rax,rcx                     ; Inc pointer into the new data
-        movq        mm1, [rbx]                  ; Copy eight bytes to mm1
+        movd        mm1, [rbx]                  ; Copy four bytes to mm1
         paddd       mm7, mm0                    ; accumulate in mm7
 
 
         ; Row 2
-        movq        mm0, [rax]                  ; Copy eight bytes to mm0
+        movd        mm0, [rax]                  ; Copy four bytes to mm0
         punpcklbw   mm0, mm6                    ; unpack to higher prrcision
         punpcklbw   mm1, mm6
         psubsw      mm0, mm1                    ; A-B (low order) to MM0
         paddw       mm5, mm0                    ; accumulate differences in mm5
 
         pmaddwd     mm0, mm0                    ; square and accumulate
         add         rbx,rdx                     ; Inc pointer into ref data
         add         rax,rcx                     ; Inc pointer into the new data
-        movq        mm1, [rbx]                  ; Copy eight bytes to mm1
+        movd        mm1, [rbx]                  ; Copy four bytes to mm1
         paddd       mm7, mm0                    ; accumulate in mm7
 
         ; Row 3
-        movq        mm0, [rax]                  ; Copy eight bytes to mm0
-        punpcklbw   mm0, mm6                    ; unpack to higher prrcision
+        movd        mm0, [rax]                  ; Copy four bytes to mm0
+        punpcklbw   mm0, mm6                    ; unpack to higher precision
         punpcklbw   mm1, mm6
         psubsw      mm0, mm1                    ; A-B (low order) to MM0
         paddw       mm5, mm0                    ; accumulate differences in mm5
 
         pmaddwd     mm0, mm0                    ; square and accumulate
         add         rbx,rdx                     ; Inc pointer into ref data
         add         rax,rcx                     ; Inc pointer into the new data
-        movq        mm1, [rbx]                  ; Copy eight bytes to mm1
+        movd        mm1, [rbx]                  ; Copy four bytes to mm1
         paddd       mm7, mm0                    ; accumulate in mm7
 
         ; Row 4
-        movq        mm0, [rax]                  ; Copy eight bytes to mm0
+        movd        mm0, [rax]                  ; Copy four bytes to mm0
 
         punpcklbw   mm0, mm6                    ; unpack to higher prrcision
         punpcklbw   mm1, mm6
         psubsw      mm0, mm1                    ; A-B (low order) to MM0
 
         paddw       mm5, mm0                    ; accumulate differences in mm5
 
         pmaddwd     mm0, mm0                    ; square and accumulate
         paddd       mm7, mm0                    ; accumulate in mm7
 
@@ skipping 446 matching lines @@
     UNSHADOW_ARGS
     pop         rbp
     ret
 
 
 SECTION_RODATA
 ;short mmx_bi_rd[4] = { 64, 64, 64, 64};
 align 16
 mmx_bi_rd:
     times 4 dw 64