Unify subtract function used in VP8/9

This commit replaces the vp8_ prefixed subtract function with the
common vpx_subtract_block function. It removes redundant SIMD
optimization codes and unit tests.

Change-Id: I42e086c32c93c6125e452dcaa6ed04337fe028d9

7 files changed, 24 insertions, 711 deletions

diff --git a/vp8/encoder/arm/neon/subtract_neon.c b/vp8/encoder/arm/neon/subtract_neon.c
deleted file mode 100644
index d3ab7b165..000000000
--- a/vp8/encoder/arm/neon/subtract_neon.c
+++ /dev/null
@@ -1,154 +0,0 @@
-/*
- *  Copyright (c) 2014 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.
- */
-
-#include <arm_neon.h>
-#include "vp8/encoder/block.h"
-
-void vp8_subtract_b_neon(
-        BLOCK *be,
-        BLOCKD *bd,
-        int pitch) {
-    unsigned char *src_ptr, *predictor;
-    int src_stride;
-    int16_t *src_diff;
-    uint8x8_t d0u8, d1u8, d2u8, d3u8, d4u8, d5u8, d6u8, d7u8;
-    uint16x8_t q10u16, q11u16, q12u16, q13u16;
-
-    src_ptr = *be->base_src + be->src;
-    src_stride = be->src_stride;
-    predictor = bd->predictor;
-
-    d0u8 = vld1_u8(src_ptr);
-    src_ptr += src_stride;
-    d2u8 = vld1_u8(src_ptr);
-    src_ptr += src_stride;
-    d4u8 = vld1_u8(src_ptr);
-    src_ptr += src_stride;
-    d6u8 = vld1_u8(src_ptr);
-
-    d1u8 = vld1_u8(predictor);
-    predictor += pitch;
-    d3u8 = vld1_u8(predictor);
-    predictor += pitch;
-    d5u8 = vld1_u8(predictor);
-    predictor += pitch;
-    d7u8 = vld1_u8(predictor);
-
-    q10u16 = vsubl_u8(d0u8, d1u8);
-    q11u16 = vsubl_u8(d2u8, d3u8);
-    q12u16 = vsubl_u8(d4u8, d5u8);
-    q13u16 = vsubl_u8(d6u8, d7u8);
-
-    src_diff = be->src_diff;
-    vst1_u16((uint16_t *)src_diff, vget_low_u16(q10u16));
-    src_diff += pitch;
-    vst1_u16((uint16_t *)src_diff, vget_low_u16(q11u16));
-    src_diff += pitch;
-    vst1_u16((uint16_t *)src_diff, vget_low_u16(q12u16));
-    src_diff += pitch;
-    vst1_u16((uint16_t *)src_diff, vget_low_u16(q13u16));
-    return;
-}
-
-void vp8_subtract_mby_neon(
-        int16_t *diff,
-        unsigned char *src,
-        int src_stride,
-        unsigned char *pred,
-        int pred_stride) {
-    int i;
-    uint8x16_t q0u8, q1u8, q2u8, q3u8;
-    uint16x8_t q8u16, q9u16, q10u16, q11u16;
-
-    for (i = 0; i < 8; i++) {  // subtract_mby_loop
-        q0u8 = vld1q_u8(src);
-        src += src_stride;
-        q2u8 = vld1q_u8(src);
-        src += src_stride;
-        q1u8 = vld1q_u8(pred);
-        pred += pred_stride;
-        q3u8 = vld1q_u8(pred);
-        pred += pred_stride;
-
-        q8u16 = vsubl_u8(vget_low_u8(q0u8), vget_low_u8(q1u8));
-        q9u16 = vsubl_u8(vget_high_u8(q0u8), vget_high_u8(q1u8));
-        q10u16 = vsubl_u8(vget_low_u8(q2u8), vget_low_u8(q3u8));
-        q11u16 = vsubl_u8(vget_high_u8(q2u8), vget_high_u8(q3u8));
-
-        vst1q_u16((uint16_t *)diff, q8u16);
-        diff += 8;
-        vst1q_u16((uint16_t *)diff, q9u16);
-        diff += 8;
-        vst1q_u16((uint16_t *)diff, q10u16);
-        diff += 8;
-        vst1q_u16((uint16_t *)diff, q11u16);
-        diff += 8;
-    }
-    return;
-}
-
-void vp8_subtract_mbuv_neon(
-        int16_t *diff,
-        unsigned char *usrc,
-        unsigned char *vsrc,
-        int src_stride,
-        unsigned char *upred,
-        unsigned char *vpred,
-        int pred_stride) {
-    int i, j;
-    unsigned char *src_ptr, *pred_ptr;
-    uint8x8_t d0u8, d1u8, d2u8, d3u8, d4u8, d5u8, d6u8, d7u8;
-    uint16x8_t q8u16, q9u16, q10u16, q11u16;
-
-    diff += 256;
-    for (i = 0; i < 2; i++) {
-        if (i == 0) {
-            src_ptr = usrc;
-            pred_ptr = upred;
-        } else if (i == 1) {
-            src_ptr = vsrc;
-            pred_ptr = vpred;
-        }
-
-        for (j = 0; j < 2; j++) {
-            d0u8 = vld1_u8(src_ptr);
-            src_ptr += src_stride;
-            d1u8 = vld1_u8(pred_ptr);
-            pred_ptr += pred_stride;
-            d2u8 = vld1_u8(src_ptr);
-            src_ptr += src_stride;
-            d3u8 = vld1_u8(pred_ptr);
-            pred_ptr += pred_stride;
-            d4u8 = vld1_u8(src_ptr);
-            src_ptr += src_stride;
-            d5u8 = vld1_u8(pred_ptr);
-            pred_ptr += pred_stride;
-            d6u8 = vld1_u8(src_ptr);
-            src_ptr += src_stride;
-            d7u8 = vld1_u8(pred_ptr);
-            pred_ptr += pred_stride;
-
-            q8u16  = vsubl_u8(d0u8, d1u8);
-            q9u16  = vsubl_u8(d2u8, d3u8);
-            q10u16 = vsubl_u8(d4u8, d5u8);
-            q11u16 = vsubl_u8(d6u8, d7u8);
-
-            vst1q_u16((uint16_t *)diff, q8u16);
-            diff += 8;
-            vst1q_u16((uint16_t *)diff, q9u16);
-            diff += 8;
-            vst1q_u16((uint16_t *)diff, q10u16);
-            diff += 8;
-            vst1q_u16((uint16_t *)diff, q11u16);
-            diff += 8;
-        }
-    }
-    return;
-}
diff --git a/vp8/encoder/encodemb.c b/vp8/encoder/encodemb.c
index 820b1376b..cf180c12d 100644
--- a/vp8/encoder/encodemb.c
+++ b/vp8/encoder/encodemb.c
@@ -8,6 +8,7 @@
  *  be found in the AUTHORS file in the root of the source tree.
  */
 
+#include "./vpx_dsp_rtcd.h"
 
 #include "vpx_config.h"
 #include "vp8_rtcd.h"
@@ -19,80 +20,29 @@
 #include "vpx_mem/vpx_mem.h"
 #include "rdopt.h"
 
-// TODO(jingning,johannkoenig): use vpx_subtract_block to replace
-// codec specified vp9_subtract_ functions.
-void vp8_subtract_b_c(BLOCK *be, BLOCKD *bd, int pitch)
-{
-    unsigned char *src_ptr = (*(be->base_src) + be->src);
-    short *diff_ptr = be->src_diff;
-    unsigned char *pred_ptr = bd->predictor;
-    int src_stride = be->src_stride;
-
-    int r, c;
+void vp8_subtract_b(BLOCK *be, BLOCKD *bd, int pitch) {
+  unsigned char *src_ptr = (*(be->base_src) + be->src);
+  short *diff_ptr = be->src_diff;
+  unsigned char *pred_ptr = bd->predictor;
+  int src_stride = be->src_stride;
 
-    for (r = 0; r < 4; r++)
-    {
-        for (c = 0; c < 4; c++)
-        {
-            diff_ptr[c] = src_ptr[c] - pred_ptr[c];
-        }
-
-        diff_ptr += pitch;
-        pred_ptr += pitch;
-        src_ptr  += src_stride;
-    }
+  vpx_subtract_block(4, 4, diff_ptr, pitch, src_ptr, src_stride,
+                     pred_ptr, pitch);
 }
 
-void vp8_subtract_mbuv_c(short *diff, unsigned char *usrc, unsigned char *vsrc,
+void vp8_subtract_mbuv(short *diff, unsigned char *usrc, unsigned char *vsrc,
                          int src_stride, unsigned char *upred,
-                         unsigned char *vpred, int pred_stride)
-{
-    short *udiff = diff + 256;
-    short *vdiff = diff + 320;
-
-    int r, c;
+                         unsigned char *vpred, int pred_stride) {
+  short *udiff = diff + 256;
+  short *vdiff = diff + 320;
 
-    for (r = 0; r < 8; r++)
-    {
-        for (c = 0; c < 8; c++)
-        {
-            udiff[c] = usrc[c] - upred[c];
-        }
-
-        udiff += 8;
-        upred += pred_stride;
-        usrc  += src_stride;
-    }
-
-    for (r = 0; r < 8; r++)
-    {
-        for (c = 0; c < 8; c++)
-        {
-            vdiff[c] = vsrc[c] - vpred[c];
-        }
-
-        vdiff += 8;
-        vpred += pred_stride;
-        vsrc  += src_stride;
-    }
+  vpx_subtract_block(8, 8, udiff, 8, usrc, src_stride, upred, pred_stride);
+  vpx_subtract_block(8, 8, vdiff, 8, vsrc, src_stride, vpred, pred_stride);
 }
 
-void vp8_subtract_mby_c(short *diff, unsigned char *src, int src_stride,
-                        unsigned char *pred, int pred_stride)
-{
-    int r, c;
-
-    for (r = 0; r < 16; r++)
-    {
-        for (c = 0; c < 16; c++)
-        {
-            diff[c] = src[c] - pred[c];
-        }
-
-        diff += 16;
-        pred += pred_stride;
-        src  += src_stride;
-    }
+void vp8_subtract_mby(short *diff, unsigned char *src, int src_stride,
+                      unsigned char *pred, int pred_stride) {
+  vpx_subtract_block(16, 16, diff, 16, src, src_stride, pred, pred_stride);
 }
 
 static void vp8_subtract_mb(MACROBLOCK *x)
diff --git a/vp8/encoder/encodemb.h b/vp8/encoder/encodemb.h
index 0b3ec875e..10b3d8651 100644
--- a/vp8/encoder/encodemb.h
+++ b/vp8/encoder/encodemb.h
@@ -19,6 +19,13 @@ extern "C" {
 #endif
 void vp8_encode_inter16x16(MACROBLOCK *x);
 
+void vp8_subtract_b(BLOCK *be, BLOCKD *bd, int pitch);
+void vp8_subtract_mbuv(short *diff, unsigned char *usrc, unsigned char *vsrc,
+                       int src_stride, unsigned char *upred,
+                       unsigned char *vpred, int pred_stride);
+void vp8_subtract_mby(short *diff, unsigned char *src, int src_stride,
+                      unsigned char *pred, int pred_stride);
+
 void vp8_build_dcblock(MACROBLOCK *b);
 void vp8_transform_mb(MACROBLOCK *mb);
 void vp8_transform_mbuv(MACROBLOCK *x);
diff --git a/vp8/encoder/x86/subtract_mmx.asm b/vp8/encoder/x86/subtract_mmx.asm
deleted file mode 100644
index 794dd22f7..000000000
--- a/vp8/encoder/x86/subtract_mmx.asm
+++ /dev/null
@@ -1,223 +0,0 @@
-;
-;  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.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;void vp8_subtract_b_mmx_impl(unsigned char *z,  int src_stride,
-;                            short *diff, unsigned char *Predictor,
-;                            int pitch);
-global sym(vp8_subtract_b_mmx_impl) PRIVATE
-sym(vp8_subtract_b_mmx_impl):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 5
-    push rsi
-    push rdi
-    ; end prolog
-
-
-        mov     rdi,        arg(2) ;diff
-        mov     rax,        arg(3) ;Predictor
-        mov     rsi,        arg(0) ;z
-        movsxd  rdx,        dword ptr arg(1);src_stride;
-        movsxd  rcx,        dword ptr arg(4);pitch
-        pxor    mm7,        mm7
-
-        movd    mm0,        [rsi]
-        movd    mm1,        [rax]
-        punpcklbw   mm0,    mm7
-        punpcklbw   mm1,    mm7
-        psubw   mm0,        mm1
-        movq    [rdi],      mm0
-
-
-        movd    mm0,        [rsi+rdx]
-        movd    mm1,        [rax+rcx]
-        punpcklbw   mm0,    mm7
-        punpcklbw   mm1,    mm7
-        psubw   mm0,        mm1
-        movq    [rdi+rcx*2],mm0
-
-
-        movd    mm0,        [rsi+rdx*2]
-        movd    mm1,        [rax+rcx*2]
-        punpcklbw   mm0,    mm7
-        punpcklbw   mm1,    mm7
-        psubw   mm0,        mm1
-        movq    [rdi+rcx*4],        mm0
-
-        lea     rsi,        [rsi+rdx*2]
-        lea     rcx,        [rcx+rcx*2]
-
-
-
-        movd    mm0,        [rsi+rdx]
-        movd    mm1,        [rax+rcx]
-        punpcklbw   mm0,    mm7
-        punpcklbw   mm1,    mm7
-        psubw   mm0,        mm1
-        movq    [rdi+rcx*2],        mm0
-
-    ; begin epilog
-    pop rdi
-    pop rsi
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
-;void vp8_subtract_mby_mmx(short *diff, unsigned char *src, int src_stride,
-;unsigned char *pred, int pred_stride)
-global sym(vp8_subtract_mby_mmx) PRIVATE
-sym(vp8_subtract_mby_mmx):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 5
-    push rsi
-    push rdi
-    ; end prolog
-
-    mov         rdi,        arg(0)          ;diff
-    mov         rsi,        arg(1)          ;src
-    movsxd      rdx,        dword ptr arg(2);src_stride
-    mov         rax,        arg(3)          ;pred
-    push        rbx
-    movsxd      rbx,        dword ptr arg(4);pred_stride
-
-    pxor        mm0,        mm0
-    mov         rcx,        16
-
-
-.submby_loop:
-    movq        mm1,        [rsi]
-    movq        mm3,        [rax]
-
-    movq        mm2,        mm1
-    movq        mm4,        mm3
-
-    punpcklbw   mm1,        mm0
-    punpcklbw   mm3,        mm0
-
-    punpckhbw   mm2,        mm0
-    punpckhbw   mm4,        mm0
-
-    psubw       mm1,        mm3
-    psubw       mm2,        mm4
-
-    movq        [rdi],      mm1
-    movq        [rdi+8],    mm2
-
-    movq        mm1,        [rsi+8]
-    movq        mm3,        [rax+8]
-
-    movq        mm2,        mm1
-    movq        mm4,        mm3
-
-    punpcklbw   mm1,        mm0
-    punpcklbw   mm3,        mm0
-
-    punpckhbw   mm2,        mm0
-    punpckhbw   mm4,        mm0
-
-    psubw       mm1,        mm3
-    psubw       mm2,        mm4
-
-    movq        [rdi+16],   mm1
-    movq        [rdi+24],   mm2
-    add         rdi,        32
-    lea         rax,        [rax+rbx]
-    lea         rsi,        [rsi+rdx]
-    dec         rcx
-    jnz         .submby_loop
-
-    pop rbx
-    pop rdi
-    pop rsi
-    ; begin epilog
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
-
-;vp8_subtract_mbuv_mmx(short *diff, unsigned char *usrc, unsigned char *vsrc,
-;                         int src_stride, unsigned char *upred,
-;                         unsigned char *vpred, int pred_stride)
-
-global sym(vp8_subtract_mbuv_mmx) PRIVATE
-sym(vp8_subtract_mbuv_mmx):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 7
-    push rsi
-    push rdi
-    ; end prolog
-
-    mov         rdi,        arg(0)          ;diff
-    mov         rsi,        arg(1)          ;usrc
-    movsxd      rdx,        dword ptr arg(3);src_stride;
-    mov         rax,        arg(4)          ;upred
-    add         rdi,        256*2           ;diff = diff + 256 (shorts)
-    mov         rcx,        8
-    push        rbx
-    movsxd      rbx,        dword ptr arg(6);pred_stride
-
-    pxor        mm7,        mm7
-
-.submbu_loop:
-    movq        mm0,        [rsi]
-    movq        mm1,        [rax]
-    movq        mm3,        mm0
-    movq        mm4,        mm1
-    punpcklbw   mm0,        mm7
-    punpcklbw   mm1,        mm7
-    punpckhbw   mm3,        mm7
-    punpckhbw   mm4,        mm7
-    psubw       mm0,        mm1
-    psubw       mm3,        mm4
-    movq        [rdi],      mm0
-    movq        [rdi+8],    mm3
-    add         rdi, 16
-    add         rsi, rdx
-    add         rax, rbx
-
-    dec         rcx
-    jnz         .submbu_loop
-
-    mov         rsi,        arg(2)          ;vsrc
-    mov         rax,        arg(5)          ;vpred
-    mov         rcx,        8
-
-.submbv_loop:
-    movq        mm0,        [rsi]
-    movq        mm1,        [rax]
-    movq        mm3,        mm0
-    movq        mm4,        mm1
-    punpcklbw   mm0,        mm7
-    punpcklbw   mm1,        mm7
-    punpckhbw   mm3,        mm7
-    punpckhbw   mm4,        mm7
-    psubw       mm0,        mm1
-    psubw       mm3,        mm4
-    movq        [rdi],      mm0
-    movq        [rdi+8],    mm3
-    add         rdi, 16
-    add         rsi, rdx
-    add         rax, rbx
-
-    dec         rcx
-    jnz         .submbv_loop
-
-    pop         rbx
-    ; begin epilog
-    pop rdi
-    pop rsi
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
diff --git a/vp8/encoder/x86/subtract_sse2.asm b/vp8/encoder/x86/subtract_sse2.asm
deleted file mode 100644
index a5d17f5be..000000000
--- a/vp8/encoder/x86/subtract_sse2.asm
+++ /dev/null
@@ -1,245 +0,0 @@
-;
-;  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.
-;
-
-
-%include "vpx_ports/x86_abi_support.asm"
-
-;void vp8_subtract_b_sse2_impl(unsigned char *z,  int src_stride,
-;                            short *diff, unsigned char *Predictor,
-;                            int pitch);
-global sym(vp8_subtract_b_sse2_impl) PRIVATE
-sym(vp8_subtract_b_sse2_impl):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 5
-    GET_GOT     rbx
-    push rsi
-    push rdi
-    ; end prolog
-
-        mov     rdi,        arg(2) ;diff
-        mov     rax,        arg(3) ;Predictor
-        mov     rsi,        arg(0) ;z
-        movsxd  rdx,        dword ptr arg(1);src_stride;
-        movsxd  rcx,        dword ptr arg(4);pitch
-        pxor    mm7,        mm7
-
-        movd    mm0,        [rsi]
-        movd    mm1,        [rax]
-        punpcklbw   mm0,    mm7
-        punpcklbw   mm1,    mm7
-        psubw   mm0,        mm1
-        movq    MMWORD PTR [rdi],      mm0
-
-        movd    mm0,        [rsi+rdx]
-        movd    mm1,        [rax+rcx]
-        punpcklbw   mm0,    mm7
-        punpcklbw   mm1,    mm7
-        psubw   mm0,        mm1
-        movq    MMWORD PTR [rdi+rcx*2], mm0
-
-        movd    mm0,        [rsi+rdx*2]
-        movd    mm1,        [rax+rcx*2]
-        punpcklbw   mm0,    mm7
-        punpcklbw   mm1,    mm7
-        psubw   mm0,        mm1
-        movq    MMWORD PTR [rdi+rcx*4], mm0
-
-        lea     rsi,        [rsi+rdx*2]
-        lea     rcx,        [rcx+rcx*2]
-
-        movd    mm0,        [rsi+rdx]
-        movd    mm1,        [rax+rcx]
-        punpcklbw   mm0,    mm7
-        punpcklbw   mm1,    mm7
-        psubw   mm0,        mm1
-        movq    MMWORD PTR [rdi+rcx*2], mm0
-
-    ; begin epilog
-    pop rdi
-    pop rsi
-    RESTORE_GOT
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
-
-;void vp8_subtract_mby_sse2(short *diff, unsigned char *src, int src_stride,
-;unsigned char *pred, int pred_stride)
-global sym(vp8_subtract_mby_sse2) PRIVATE
-sym(vp8_subtract_mby_sse2):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 5
-    GET_GOT     rbx
-    push rsi
-    push rdi
-    ; end prolog
-
-    mov         rdi,        arg(0)          ;diff
-    mov         rsi,        arg(1)          ;src
-    movsxd      rdx,        dword ptr arg(2);src_stride
-    mov         rax,        arg(3)          ;pred
-    movdqa      xmm4,       [GLOBAL(t80)]
-    push        rbx
-    mov         rcx,        8               ; do two lines at one time
-    movsxd      rbx,        dword ptr arg(4);pred_stride
-
-.submby_loop:
-    movdqa      xmm0,       [rsi]           ; src
-    movdqa      xmm1,       [rax]           ; pred
-
-    movdqa      xmm2,       xmm0
-    psubb       xmm0,       xmm1
-
-    pxor        xmm1,       xmm4            ;convert to signed values
-    pxor        xmm2,       xmm4
-    pcmpgtb     xmm1,       xmm2            ; obtain sign information
-
-    movdqa      xmm2,       xmm0
-    punpcklbw   xmm0,       xmm1            ; put sign back to subtraction
-    punpckhbw   xmm2,       xmm1            ; put sign back to subtraction
-
-    movdqa      xmm3,       [rsi + rdx]
-    movdqa      xmm5,       [rax + rbx]
-
-    lea         rsi,        [rsi+rdx*2]
-    lea         rax,        [rax+rbx*2]
-
-    movdqa      [rdi],      xmm0
-    movdqa      [rdi +16],  xmm2
-
-    movdqa      xmm1,       xmm3
-    psubb       xmm3,       xmm5
-
-    pxor        xmm5,       xmm4            ;convert to signed values
-    pxor        xmm1,       xmm4
-    pcmpgtb     xmm5,       xmm1            ; obtain sign information
-
-    movdqa      xmm1,       xmm3
-    punpcklbw   xmm3,       xmm5            ; put sign back to subtraction
-    punpckhbw   xmm1,       xmm5            ; put sign back to subtraction
-
-    movdqa      [rdi +32],  xmm3
-    movdqa      [rdi +48],  xmm1
-
-    add         rdi,        64
-    dec         rcx
-    jnz         .submby_loop
-
-    pop rbx
-    pop rdi
-    pop rsi
-    ; begin epilog
-    RESTORE_GOT
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
-;vp8_subtract_mbuv_sse2(short *diff, unsigned char *usrc, unsigned char *vsrc,
-;                         int src_stride, unsigned char *upred,
-;                         unsigned char *vpred, int pred_stride)
-global sym(vp8_subtract_mbuv_sse2) PRIVATE
-sym(vp8_subtract_mbuv_sse2):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 7
-    GET_GOT     rbx
-    push rsi
-    push rdi
-    ; end prolog
-
-    movdqa      xmm4,       [GLOBAL(t80)]
-    mov         rdi,        arg(0)          ;diff
-    mov         rsi,        arg(1)          ;usrc
-    movsxd      rdx,        dword ptr arg(3);src_stride;
-    mov         rax,        arg(4)          ;upred
-    add         rdi,        256*2           ;diff = diff + 256 (shorts)
-    mov         rcx,        4
-    push        rbx
-    movsxd      rbx,        dword ptr arg(6);pred_stride
-
-    ;u
-.submbu_loop:
-    movq        xmm0,       [rsi]           ; src
-    movq        xmm2,       [rsi+rdx]       ; src -- next line
-    movq        xmm1,       [rax]           ; pred
-    movq        xmm3,       [rax+rbx]       ; pred -- next line
-    lea         rsi,        [rsi + rdx*2]
-    lea         rax,        [rax + rbx*2]
-
-    punpcklqdq  xmm0,       xmm2
-    punpcklqdq  xmm1,       xmm3
-
-    movdqa      xmm2,       xmm0
-    psubb       xmm0,       xmm1            ; subtraction with sign missed
-
-    pxor        xmm1,       xmm4            ;convert to signed values
-    pxor        xmm2,       xmm4
-    pcmpgtb     xmm1,       xmm2            ; obtain sign information
-
-    movdqa      xmm2,       xmm0
-    movdqa      xmm3,       xmm1
-    punpcklbw   xmm0,       xmm1            ; put sign back to subtraction
-    punpckhbw   xmm2,       xmm3            ; put sign back to subtraction
-
-    movdqa      [rdi],      xmm0            ; store difference
-    movdqa      [rdi +16],  xmm2            ; store difference
-    add         rdi,        32
-    sub         rcx, 1
-    jnz         .submbu_loop
-
-    mov         rsi,        arg(2)          ;vsrc
-    mov         rax,        arg(5)          ;vpred
-    mov         rcx,        4
-
-    ;v
-.submbv_loop:
-    movq        xmm0,       [rsi]           ; src
-    movq        xmm2,       [rsi+rdx]       ; src -- next line
-    movq        xmm1,       [rax]           ; pred
-    movq        xmm3,       [rax+rbx]       ; pred -- next line
-    lea         rsi,        [rsi + rdx*2]
-    lea         rax,        [rax + rbx*2]
-
-    punpcklqdq  xmm0,       xmm2
-    punpcklqdq  xmm1,       xmm3
-
-    movdqa      xmm2,       xmm0
-    psubb       xmm0,       xmm1            ; subtraction with sign missed
-
-    pxor        xmm1,       xmm4            ;convert to signed values
-    pxor        xmm2,       xmm4
-    pcmpgtb     xmm1,       xmm2            ; obtain sign information
-
-    movdqa      xmm2,       xmm0
-    movdqa      xmm3,       xmm1
-    punpcklbw   xmm0,       xmm1            ; put sign back to subtraction
-    punpckhbw   xmm2,       xmm3            ; put sign back to subtraction
-
-    movdqa      [rdi],      xmm0            ; store difference
-    movdqa      [rdi +16],  xmm2            ; store difference
-    add         rdi,        32
-    sub         rcx, 1
-    jnz         .submbv_loop
-
-    pop         rbx
-    ; begin epilog
-    pop rdi
-    pop rsi
-    RESTORE_GOT
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
-SECTION_RODATA
-align 16
-t80:
-    times 16 db 0x80
diff --git a/vp8/encoder/x86/vp8_enc_stubs_mmx.c b/vp8/encoder/x86/vp8_enc_stubs_mmx.c
index cf3d8ca4a..7bf5155c9 100644
--- a/vp8/encoder/x86/vp8_enc_stubs_mmx.c
+++ b/vp8/encoder/x86/vp8_enc_stubs_mmx.c
@@ -65,14 +65,3 @@ int vp8_mbuverror_mmx(MACROBLOCK *mb)
     return vp8_mbuverror_mmx_impl(s_ptr, d_ptr);
 }
 
-void vp8_subtract_b_mmx_impl(unsigned char *z,  int src_stride,
-                             short *diff, unsigned char *predictor,
-                             int pitch);
-void vp8_subtract_b_mmx(BLOCK *be, BLOCKD *bd, int pitch)
-{
-    unsigned char *z = *(be->base_src) + be->src;
-    unsigned int  src_stride = be->src_stride;
-    short *diff = &be->src_diff[0];
-    unsigned char *predictor = &bd->predictor[0];
-    vp8_subtract_b_mmx_impl(z, src_stride, diff, predictor, pitch);
-}
diff --git a/vp8/encoder/x86/vp8_enc_stubs_sse2.c b/vp8/encoder/x86/vp8_enc_stubs_sse2.c
index 3dfbee368..be9aaf3c9 100644
--- a/vp8/encoder/x86/vp8_enc_stubs_sse2.c
+++ b/vp8/encoder/x86/vp8_enc_stubs_sse2.c
@@ -30,14 +30,3 @@ int vp8_mbuverror_xmm(MACROBLOCK *mb)
     return vp8_mbuverror_xmm_impl(s_ptr, d_ptr);
 }
 
-void vp8_subtract_b_sse2_impl(unsigned char *z,  int src_stride,
-                             short *diff, unsigned char *predictor,
-                             int pitch);
-void vp8_subtract_b_sse2(BLOCK *be, BLOCKD *bd, int pitch)
-{
-    unsigned char *z = *(be->base_src) + be->src;
-    unsigned int  src_stride = be->src_stride;
-    short *diff = &be->src_diff[0];
-    unsigned char *predictor = &bd->predictor[0];
-    vp8_subtract_b_sse2_impl(z, src_stride, diff, predictor, pitch);
-}