Move variance functions to vpx_dsp

subpel functions will be moved in another patch.

Change-Id: Idb2e049bad0b9b32ac42cc7731cd6903de2826ce

24 files changed, 80 insertions, 2287 deletions

diff --git a/vp8/common/arm/armv6/vp8_variance16x16_armv6.asm b/vp8/common/arm/armv6/vp8_variance16x16_armv6.asm
deleted file mode 100644
index 39919579f..000000000
--- a/vp8/common/arm/armv6/vp8_variance16x16_armv6.asm
+++ /dev/null
@@ -1,154 +0,0 @@
-;
-;  Copyright (c) 2011 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.
-;
-
-
-    EXPORT  |vp8_variance16x16_armv6|
-
-    ARM
-    REQUIRE8
-    PRESERVE8
-
-    AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0    unsigned char *src_ptr
-; r1    int source_stride
-; r2    unsigned char *ref_ptr
-; r3    int  recon_stride
-; stack unsigned int *sse
-|vp8_variance16x16_armv6| PROC
-
-    stmfd   sp!, {r4-r12, lr}
-
-    pld     [r0, r1, lsl #0]
-    pld     [r2, r3, lsl #0]
-
-    mov     r8, #0              ; initialize sum = 0
-    mov     r11, #0             ; initialize sse = 0
-    mov     r12, #16            ; set loop counter to 16 (=block height)
-
-loop
-    ; 1st 4 pixels
-    ldr     r4, [r0, #0]        ; load 4 src pixels
-    ldr     r5, [r2, #0]        ; load 4 ref pixels
-
-    mov     lr, #0              ; constant zero
-
-    usub8   r6, r4, r5          ; calculate difference
-    pld     [r0, r1, lsl #1]
-    sel     r7, r6, lr          ; select bytes with positive difference
-    usub8   r9, r5, r4          ; calculate difference with reversed operands
-    pld     [r2, r3, lsl #1]
-    sel     r6, r9, lr          ; select bytes with negative difference
-
-    ; calculate partial sums
-    usad8   r4, r7, lr          ; calculate sum of positive differences
-    usad8   r5, r6, lr          ; calculate sum of negative differences
-    orr     r6, r6, r7          ; differences of all 4 pixels
-    ; calculate total sum
-    adds    r8, r8, r4          ; add positive differences to sum
-    subs    r8, r8, r5          ; subtract negative differences from sum
-
-    ; calculate sse
-    uxtb16  r5, r6              ; byte (two pixels) to halfwords
-    uxtb16  r10, r6, ror #8     ; another two pixels to halfwords
-    smlad   r11, r5, r5, r11    ; dual signed multiply, add and accumulate (1)
-
-    ; 2nd 4 pixels
-    ldr     r4, [r0, #4]        ; load 4 src pixels
-    ldr     r5, [r2, #4]        ; load 4 ref pixels
-    smlad   r11, r10, r10, r11  ; dual signed multiply, add and accumulate (2)
-
-    usub8   r6, r4, r5          ; calculate difference
-    sel     r7, r6, lr          ; select bytes with positive difference
-    usub8   r9, r5, r4          ; calculate difference with reversed operands
-    sel     r6, r9, lr          ; select bytes with negative difference
-
-    ; calculate partial sums
-    usad8   r4, r7, lr          ; calculate sum of positive differences
-    usad8   r5, r6, lr          ; calculate sum of negative differences
-    orr     r6, r6, r7          ; differences of all 4 pixels
-
-    ; calculate total sum
-    add     r8, r8, r4          ; add positive differences to sum
-    sub     r8, r8, r5          ; subtract negative differences from sum
-
-    ; calculate sse
-    uxtb16  r5, r6              ; byte (two pixels) to halfwords
-    uxtb16  r10, r6, ror #8     ; another two pixels to halfwords
-    smlad   r11, r5, r5, r11    ; dual signed multiply, add and accumulate (1)
-
-    ; 3rd 4 pixels
-    ldr     r4, [r0, #8]        ; load 4 src pixels
-    ldr     r5, [r2, #8]        ; load 4 ref pixels
-    smlad   r11, r10, r10, r11  ; dual signed multiply, add and accumulate (2)
-
-    usub8   r6, r4, r5          ; calculate difference
-    sel     r7, r6, lr          ; select bytes with positive difference
-    usub8   r9, r5, r4          ; calculate difference with reversed operands
-    sel     r6, r9, lr          ; select bytes with negative difference
-
-    ; calculate partial sums
-    usad8   r4, r7, lr          ; calculate sum of positive differences
-    usad8   r5, r6, lr          ; calculate sum of negative differences
-    orr     r6, r6, r7          ; differences of all 4 pixels
-
-    ; calculate total sum
-    add     r8, r8, r4          ; add positive differences to sum
-    sub     r8, r8, r5          ; subtract negative differences from sum
-
-    ; calculate sse
-    uxtb16  r5, r6              ; byte (two pixels) to halfwords
-    uxtb16  r10, r6, ror #8     ; another two pixels to halfwords
-    smlad   r11, r5, r5, r11    ; dual signed multiply, add and accumulate (1)
-
-    ; 4th 4 pixels
-    ldr     r4, [r0, #12]       ; load 4 src pixels
-    ldr     r5, [r2, #12]       ; load 4 ref pixels
-    smlad   r11, r10, r10, r11  ; dual signed multiply, add and accumulate (2)
-
-    usub8   r6, r4, r5          ; calculate difference
-    add     r0, r0, r1          ; set src_ptr to next row
-    sel     r7, r6, lr          ; select bytes with positive difference
-    usub8   r9, r5, r4          ; calculate difference with reversed operands
-    add     r2, r2, r3          ; set dst_ptr to next row
-    sel     r6, r9, lr          ; select bytes with negative difference
-
-    ; calculate partial sums
-    usad8   r4, r7, lr          ; calculate sum of positive differences
-    usad8   r5, r6, lr          ; calculate sum of negative differences
-    orr     r6, r6, r7          ; differences of all 4 pixels
-
-    ; calculate total sum
-    add     r8, r8, r4          ; add positive differences to sum
-    sub     r8, r8, r5          ; subtract negative differences from sum
-
-    ; calculate sse
-    uxtb16  r5, r6              ; byte (two pixels) to halfwords
-    uxtb16  r10, r6, ror #8     ; another two pixels to halfwords
-    smlad   r11, r5, r5, r11    ; dual signed multiply, add and accumulate (1)
-    smlad   r11, r10, r10, r11  ; dual signed multiply, add and accumulate (2)
-
-
-    subs    r12, r12, #1
-
-    bne     loop
-
-    ; return stuff
-    ldr     r6, [sp, #40]       ; get address of sse
-    mul     r0, r8, r8          ; sum * sum
-    str     r11, [r6]           ; store sse
-    sub     r0, r11, r0, lsr #8 ; return (sse - ((sum * sum) >> 8))
-
-    ldmfd   sp!, {r4-r12, pc}
-
-    ENDP
-
-    END
-
diff --git a/vp8/common/arm/armv6/vp8_variance8x8_armv6.asm b/vp8/common/arm/armv6/vp8_variance8x8_armv6.asm
deleted file mode 100644
index 915ee4993..000000000
--- a/vp8/common/arm/armv6/vp8_variance8x8_armv6.asm
+++ /dev/null
@@ -1,101 +0,0 @@
-;
-;  Copyright (c) 2011 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.
-;
-
-
-    EXPORT  |vp8_variance8x8_armv6|
-
-    ARM
-
-    AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0    unsigned char *src_ptr
-; r1    int source_stride
-; r2    unsigned char *ref_ptr
-; r3    int  recon_stride
-; stack unsigned int *sse
-|vp8_variance8x8_armv6| PROC
-
-    push    {r4-r10, lr}
-
-    pld     [r0, r1, lsl #0]
-    pld     [r2, r3, lsl #0]
-
-    mov     r12, #8             ; set loop counter to 8 (=block height)
-    mov     r4, #0              ; initialize sum = 0
-    mov     r5, #0              ; initialize sse = 0
-
-loop
-    ; 1st 4 pixels
-    ldr     r6, [r0, #0x0]      ; load 4 src pixels
-    ldr     r7, [r2, #0x0]      ; load 4 ref pixels
-
-    mov     lr, #0              ; constant zero
-
-    usub8   r8, r6, r7          ; calculate difference
-    pld     [r0, r1, lsl #1]
-    sel     r10, r8, lr         ; select bytes with positive difference
-    usub8   r9, r7, r6          ; calculate difference with reversed operands
-    pld     [r2, r3, lsl #1]
-    sel     r8, r9, lr          ; select bytes with negative difference
-
-    ; calculate partial sums
-    usad8   r6, r10, lr         ; calculate sum of positive differences
-    usad8   r7, r8, lr          ; calculate sum of negative differences
-    orr     r8, r8, r10         ; differences of all 4 pixels
-    ; calculate total sum
-    add    r4, r4, r6           ; add positive differences to sum
-    sub    r4, r4, r7           ; subtract negative differences from sum
-
-    ; calculate sse
-    uxtb16  r7, r8              ; byte (two pixels) to halfwords
-    uxtb16  r10, r8, ror #8     ; another two pixels to halfwords
-    smlad   r5, r7, r7, r5      ; dual signed multiply, add and accumulate (1)
-
-    ; 2nd 4 pixels
-    ldr     r6, [r0, #0x4]      ; load 4 src pixels
-    ldr     r7, [r2, #0x4]      ; load 4 ref pixels
-    smlad   r5, r10, r10, r5    ; dual signed multiply, add and accumulate (2)
-
-    usub8   r8, r6, r7          ; calculate difference
-    add     r0, r0, r1          ; set src_ptr to next row
-    sel     r10, r8, lr         ; select bytes with positive difference
-    usub8   r9, r7, r6          ; calculate difference with reversed operands
-    add     r2, r2, r3          ; set dst_ptr to next row
-    sel     r8, r9, lr          ; select bytes with negative difference
-
-    ; calculate partial sums
-    usad8   r6, r10, lr         ; calculate sum of positive differences
-    usad8   r7, r8, lr          ; calculate sum of negative differences
-    orr     r8, r8, r10         ; differences of all 4 pixels
-
-    ; calculate total sum
-    add     r4, r4, r6          ; add positive differences to sum
-    sub     r4, r4, r7          ; subtract negative differences from sum
-
-    ; calculate sse
-    uxtb16  r7, r8              ; byte (two pixels) to halfwords
-    uxtb16  r10, r8, ror #8     ; another two pixels to halfwords
-    smlad   r5, r7, r7, r5      ; dual signed multiply, add and accumulate (1)
-    subs    r12, r12, #1        ; next row
-    smlad   r5, r10, r10, r5    ; dual signed multiply, add and accumulate (2)
-
-    bne     loop
-
-    ; return stuff
-    ldr     r8, [sp, #32]       ; get address of sse
-    mul     r1, r4, r4          ; sum * sum
-    str     r5, [r8]            ; store sse
-    sub     r0, r5, r1, ASR #6  ; return (sse - ((sum * sum) >> 6))
-
-    pop     {r4-r10, pc}
-
-    ENDP
-
-    END
diff --git a/vp8/common/arm/neon/variance_neon.c b/vp8/common/arm/neon/variance_neon.c
deleted file mode 100644
index 1b1979073..000000000
--- a/vp8/common/arm/neon/variance_neon.c
+++ /dev/null
@@ -1,320 +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 "vpx_ports/mem.h"
-
-unsigned int vp8_variance16x16_neon(
-        const unsigned char *src_ptr,
-        int source_stride,
-        const unsigned char *ref_ptr,
-        int recon_stride,
-        unsigned int *sse) {
-    int i;
-    int16x4_t d22s16, d23s16, d24s16, d25s16, d26s16, d27s16, d28s16, d29s16;
-    uint32x2_t d0u32, d10u32;
-    int64x1_t d0s64, d1s64;
-    uint8x16_t q0u8, q1u8, q2u8, q3u8;
-    uint16x8_t q11u16, q12u16, q13u16, q14u16;
-    int32x4_t q8s32, q9s32, q10s32;
-    int64x2_t q0s64, q1s64, q5s64;
-
-    q8s32 = vdupq_n_s32(0);
-    q9s32 = vdupq_n_s32(0);
-    q10s32 = vdupq_n_s32(0);
-
-    for (i = 0; i < 8; i++) {
-        q0u8 = vld1q_u8(src_ptr);
-        src_ptr += source_stride;
-        q1u8 = vld1q_u8(src_ptr);
-        src_ptr += source_stride;
-        __builtin_prefetch(src_ptr);
-
-        q2u8 = vld1q_u8(ref_ptr);
-        ref_ptr += recon_stride;
-        q3u8 = vld1q_u8(ref_ptr);
-        ref_ptr += recon_stride;
-        __builtin_prefetch(ref_ptr);
-
-        q11u16 = vsubl_u8(vget_low_u8(q0u8), vget_low_u8(q2u8));
-        q12u16 = vsubl_u8(vget_high_u8(q0u8), vget_high_u8(q2u8));
-        q13u16 = vsubl_u8(vget_low_u8(q1u8), vget_low_u8(q3u8));
-        q14u16 = vsubl_u8(vget_high_u8(q1u8), vget_high_u8(q3u8));
-
-        d22s16 = vreinterpret_s16_u16(vget_low_u16(q11u16));
-        d23s16 = vreinterpret_s16_u16(vget_high_u16(q11u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q11u16));
-        q9s32 = vmlal_s16(q9s32, d22s16, d22s16);
-        q10s32 = vmlal_s16(q10s32, d23s16, d23s16);
-
-        d24s16 = vreinterpret_s16_u16(vget_low_u16(q12u16));
-        d25s16 = vreinterpret_s16_u16(vget_high_u16(q12u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q12u16));
-        q9s32 = vmlal_s16(q9s32, d24s16, d24s16);
-        q10s32 = vmlal_s16(q10s32, d25s16, d25s16);
-
-        d26s16 = vreinterpret_s16_u16(vget_low_u16(q13u16));
-        d27s16 = vreinterpret_s16_u16(vget_high_u16(q13u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q13u16));
-        q9s32 = vmlal_s16(q9s32, d26s16, d26s16);
-        q10s32 = vmlal_s16(q10s32, d27s16, d27s16);
-
-        d28s16 = vreinterpret_s16_u16(vget_low_u16(q14u16));
-        d29s16 = vreinterpret_s16_u16(vget_high_u16(q14u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q14u16));
-        q9s32 = vmlal_s16(q9s32, d28s16, d28s16);
-        q10s32 = vmlal_s16(q10s32, d29s16, d29s16);
-    }
-
-    q10s32 = vaddq_s32(q10s32, q9s32);
-    q0s64 = vpaddlq_s32(q8s32);
-    q1s64 = vpaddlq_s32(q10s32);
-
-    d0s64 = vadd_s64(vget_low_s64(q0s64), vget_high_s64(q0s64));
-    d1s64 = vadd_s64(vget_low_s64(q1s64), vget_high_s64(q1s64));
-
-    q5s64 = vmull_s32(vreinterpret_s32_s64(d0s64),
-                      vreinterpret_s32_s64(d0s64));
-    vst1_lane_u32((uint32_t *)sse, vreinterpret_u32_s64(d1s64), 0);
-
-    d10u32 = vshr_n_u32(vreinterpret_u32_s64(vget_low_s64(q5s64)), 8);
-    d0u32 = vsub_u32(vreinterpret_u32_s64(d1s64), d10u32);
-
-    return vget_lane_u32(d0u32, 0);
-}
-
-unsigned int vp8_variance16x8_neon(
-        const unsigned char *src_ptr,
-        int source_stride,
-        const unsigned char *ref_ptr,
-        int recon_stride,
-        unsigned int *sse) {
-    int i;
-    int16x4_t d22s16, d23s16, d24s16, d25s16, d26s16, d27s16, d28s16, d29s16;
-    uint32x2_t d0u32, d10u32;
-    int64x1_t d0s64, d1s64;
-    uint8x16_t q0u8, q1u8, q2u8, q3u8;
-    uint16x8_t q11u16, q12u16, q13u16, q14u16;
-    int32x4_t q8s32, q9s32, q10s32;
-    int64x2_t q0s64, q1s64, q5s64;
-
-    q8s32 = vdupq_n_s32(0);
-    q9s32 = vdupq_n_s32(0);
-    q10s32 = vdupq_n_s32(0);
-
-    for (i = 0; i < 4; i++) {  // variance16x8_neon_loop
-        q0u8 = vld1q_u8(src_ptr);
-        src_ptr += source_stride;
-        q1u8 = vld1q_u8(src_ptr);
-        src_ptr += source_stride;
-        __builtin_prefetch(src_ptr);
-
-        q2u8 = vld1q_u8(ref_ptr);
-        ref_ptr += recon_stride;
-        q3u8 = vld1q_u8(ref_ptr);
-        ref_ptr += recon_stride;
-        __builtin_prefetch(ref_ptr);
-
-        q11u16 = vsubl_u8(vget_low_u8(q0u8), vget_low_u8(q2u8));
-        q12u16 = vsubl_u8(vget_high_u8(q0u8), vget_high_u8(q2u8));
-        q13u16 = vsubl_u8(vget_low_u8(q1u8), vget_low_u8(q3u8));
-        q14u16 = vsubl_u8(vget_high_u8(q1u8), vget_high_u8(q3u8));
-
-        d22s16 = vreinterpret_s16_u16(vget_low_u16(q11u16));
-        d23s16 = vreinterpret_s16_u16(vget_high_u16(q11u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q11u16));
-        q9s32 = vmlal_s16(q9s32, d22s16, d22s16);
-        q10s32 = vmlal_s16(q10s32, d23s16, d23s16);
-
-        d24s16 = vreinterpret_s16_u16(vget_low_u16(q12u16));
-        d25s16 = vreinterpret_s16_u16(vget_high_u16(q12u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q12u16));
-        q9s32 = vmlal_s16(q9s32, d24s16, d24s16);
-        q10s32 = vmlal_s16(q10s32, d25s16, d25s16);
-
-        d26s16 = vreinterpret_s16_u16(vget_low_u16(q13u16));
-        d27s16 = vreinterpret_s16_u16(vget_high_u16(q13u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q13u16));
-        q9s32 = vmlal_s16(q9s32, d26s16, d26s16);
-        q10s32 = vmlal_s16(q10s32, d27s16, d27s16);
-
-        d28s16 = vreinterpret_s16_u16(vget_low_u16(q14u16));
-        d29s16 = vreinterpret_s16_u16(vget_high_u16(q14u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q14u16));
-        q9s32 = vmlal_s16(q9s32, d28s16, d28s16);
-        q10s32 = vmlal_s16(q10s32, d29s16, d29s16);
-    }
-
-    q10s32 = vaddq_s32(q10s32, q9s32);
-    q0s64 = vpaddlq_s32(q8s32);
-    q1s64 = vpaddlq_s32(q10s32);
-
-    d0s64 = vadd_s64(vget_low_s64(q0s64), vget_high_s64(q0s64));
-    d1s64 = vadd_s64(vget_low_s64(q1s64), vget_high_s64(q1s64));
-
-    q5s64 = vmull_s32(vreinterpret_s32_s64(d0s64),
-                      vreinterpret_s32_s64(d0s64));
-    vst1_lane_u32((uint32_t *)sse, vreinterpret_u32_s64(d1s64), 0);
-
-    d10u32 = vshr_n_u32(vreinterpret_u32_s64(vget_low_s64(q5s64)), 7);
-    d0u32 = vsub_u32(vreinterpret_u32_s64(d1s64), d10u32);
-
-    return vget_lane_u32(d0u32, 0);
-}
-
-unsigned int vp8_variance8x16_neon(
-        const unsigned char *src_ptr,
-        int source_stride,
-        const unsigned char *ref_ptr,
-        int recon_stride,
-        unsigned int *sse) {
-    int i;
-    uint8x8_t d0u8, d2u8, d4u8, d6u8;
-    int16x4_t d22s16, d23s16, d24s16, d25s16;
-    uint32x2_t d0u32, d10u32;
-    int64x1_t d0s64, d1s64;
-    uint16x8_t q11u16, q12u16;
-    int32x4_t q8s32, q9s32, q10s32;
-    int64x2_t q0s64, q1s64, q5s64;
-
-    q8s32 = vdupq_n_s32(0);
-    q9s32 = vdupq_n_s32(0);
-    q10s32 = vdupq_n_s32(0);
-
-    for (i = 0; i < 8; i++) {  // variance8x16_neon_loop
-        d0u8 = vld1_u8(src_ptr);
-        src_ptr += source_stride;
-        d2u8 = vld1_u8(src_ptr);
-        src_ptr += source_stride;
-        __builtin_prefetch(src_ptr);
-
-        d4u8 = vld1_u8(ref_ptr);
-        ref_ptr += recon_stride;
-        d6u8 = vld1_u8(ref_ptr);
-        ref_ptr += recon_stride;
-        __builtin_prefetch(ref_ptr);
-
-        q11u16 = vsubl_u8(d0u8, d4u8);
-        q12u16 = vsubl_u8(d2u8, d6u8);
-
-        d22s16 = vreinterpret_s16_u16(vget_low_u16(q11u16));
-        d23s16 = vreinterpret_s16_u16(vget_high_u16(q11u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q11u16));
-        q9s32 = vmlal_s16(q9s32, d22s16, d22s16);
-        q10s32 = vmlal_s16(q10s32, d23s16, d23s16);
-
-        d24s16 = vreinterpret_s16_u16(vget_low_u16(q12u16));
-        d25s16 = vreinterpret_s16_u16(vget_high_u16(q12u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q12u16));
-        q9s32 = vmlal_s16(q9s32, d24s16, d24s16);
-        q10s32 = vmlal_s16(q10s32, d25s16, d25s16);
-    }
-
-    q10s32 = vaddq_s32(q10s32, q9s32);
-    q0s64 = vpaddlq_s32(q8s32);
-    q1s64 = vpaddlq_s32(q10s32);
-
-    d0s64 = vadd_s64(vget_low_s64(q0s64), vget_high_s64(q0s64));
-    d1s64 = vadd_s64(vget_low_s64(q1s64), vget_high_s64(q1s64));
-
-    q5s64 = vmull_s32(vreinterpret_s32_s64(d0s64),
-                      vreinterpret_s32_s64(d0s64));
-    vst1_lane_u32((uint32_t *)sse, vreinterpret_u32_s64(d1s64), 0);
-
-    d10u32 = vshr_n_u32(vreinterpret_u32_s64(vget_low_s64(q5s64)), 7);
-    d0u32 = vsub_u32(vreinterpret_u32_s64(d1s64), d10u32);
-
-    return vget_lane_u32(d0u32, 0);
-}
-
-unsigned int vp8_variance8x8_neon(
-        const unsigned char *src_ptr,
-        int source_stride,
-        const unsigned char *ref_ptr,
-        int recon_stride,
-        unsigned int *sse) {
-    int i;
-    uint8x8_t d0u8, d1u8, d2u8, d3u8, d4u8, d5u8, d6u8, d7u8;
-    int16x4_t d22s16, d23s16, d24s16, d25s16, d26s16, d27s16, d28s16, d29s16;
-    uint32x2_t d0u32, d10u32;
-    int64x1_t d0s64, d1s64;
-    uint16x8_t q11u16, q12u16, q13u16, q14u16;
-    int32x4_t q8s32, q9s32, q10s32;
-    int64x2_t q0s64, q1s64, q5s64;
-
-    q8s32 = vdupq_n_s32(0);
-    q9s32 = vdupq_n_s32(0);
-    q10s32 = vdupq_n_s32(0);
-
-    for (i = 0; i < 2; i++) {  // variance8x8_neon_loop
-        d0u8 = vld1_u8(src_ptr);
-        src_ptr += source_stride;
-        d1u8 = vld1_u8(src_ptr);
-        src_ptr += source_stride;
-        d2u8 = vld1_u8(src_ptr);
-        src_ptr += source_stride;
-        d3u8 = vld1_u8(src_ptr);
-        src_ptr += source_stride;
-
-        d4u8 = vld1_u8(ref_ptr);
-        ref_ptr += recon_stride;
-        d5u8 = vld1_u8(ref_ptr);
-        ref_ptr += recon_stride;
-        d6u8 = vld1_u8(ref_ptr);
-        ref_ptr += recon_stride;
-        d7u8 = vld1_u8(ref_ptr);
-        ref_ptr += recon_stride;
-
-        q11u16 = vsubl_u8(d0u8, d4u8);
-        q12u16 = vsubl_u8(d1u8, d5u8);
-        q13u16 = vsubl_u8(d2u8, d6u8);
-        q14u16 = vsubl_u8(d3u8, d7u8);
-
-        d22s16 = vreinterpret_s16_u16(vget_low_u16(q11u16));
-        d23s16 = vreinterpret_s16_u16(vget_high_u16(q11u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q11u16));
-        q9s32 = vmlal_s16(q9s32, d22s16, d22s16);
-        q10s32 = vmlal_s16(q10s32, d23s16, d23s16);
-
-        d24s16 = vreinterpret_s16_u16(vget_low_u16(q12u16));
-        d25s16 = vreinterpret_s16_u16(vget_high_u16(q12u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q12u16));
-        q9s32 = vmlal_s16(q9s32, d24s16, d24s16);
-        q10s32 = vmlal_s16(q10s32, d25s16, d25s16);
-
-        d26s16 = vreinterpret_s16_u16(vget_low_u16(q13u16));
-        d27s16 = vreinterpret_s16_u16(vget_high_u16(q13u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q13u16));
-        q9s32 = vmlal_s16(q9s32, d26s16, d26s16);
-        q10s32 = vmlal_s16(q10s32, d27s16, d27s16);
-
-        d28s16 = vreinterpret_s16_u16(vget_low_u16(q14u16));
-        d29s16 = vreinterpret_s16_u16(vget_high_u16(q14u16));
-        q8s32 = vpadalq_s16(q8s32, vreinterpretq_s16_u16(q14u16));
-        q9s32 = vmlal_s16(q9s32, d28s16, d28s16);
-        q10s32 = vmlal_s16(q10s32, d29s16, d29s16);
-    }
-
-    q10s32 = vaddq_s32(q10s32, q9s32);
-    q0s64 = vpaddlq_s32(q8s32);
-    q1s64 = vpaddlq_s32(q10s32);
-
-    d0s64 = vadd_s64(vget_low_s64(q0s64), vget_high_s64(q0s64));
-    d1s64 = vadd_s64(vget_low_s64(q1s64), vget_high_s64(q1s64));
-
-    q5s64 = vmull_s32(vreinterpret_s32_s64(d0s64),
-                      vreinterpret_s32_s64(d0s64));
-    vst1_lane_u32((uint32_t *)sse, vreinterpret_u32_s64(d1s64), 0);
-
-    d10u32 = vshr_n_u32(vreinterpret_u32_s64(vget_low_s64(q5s64)), 6);
-    d0u32 = vsub_u32(vreinterpret_u32_s64(d1s64), d10u32);
-
-    return vget_lane_u32(d0u32, 0);
-}
diff --git a/vp8/common/arm/variance_arm.c b/vp8/common/arm/variance_arm.c
index 467a50942..0f293f03d 100644
--- a/vp8/common/arm/variance_arm.c
+++ b/vp8/common/arm/variance_arm.c
@@ -9,10 +9,14 @@
  */
 
 #include "vpx_config.h"
-#include "vp8_rtcd.h"
+#include "./vp8_rtcd.h"
+#include "./vpx_dsp_rtcd.h"
 #include "vp8/common/variance.h"
 #include "vp8/common/filter.h"
 
+// TODO(johannkoenig): Move this to vpx_dsp or vp8/encoder
+#if CONFIG_VP8_ENCODER
+
 #if HAVE_MEDIA
 #include "vp8/common/arm/bilinearfilter_arm.h"
 
@@ -40,8 +44,8 @@ unsigned int vp8_sub_pixel_variance8x8_armv6
     vp8_filter_block2d_bil_second_pass_armv6(first_pass, second_pass,
                                              8, 8, 8, VFilter);
 
-    return vp8_variance8x8_armv6(second_pass, 8, dst_ptr,
-                                   dst_pixels_per_line, sse);
+    return vpx_variance8x8_media(second_pass, 8, dst_ptr,
+                                 dst_pixels_per_line, sse);
 }
 
 unsigned int vp8_sub_pixel_variance16x16_armv6
@@ -86,13 +90,13 @@ unsigned int vp8_sub_pixel_variance16x16_armv6
         vp8_filter_block2d_bil_second_pass_armv6(first_pass, second_pass,
                                                  16, 16, 16, VFilter);
 
-        var = vp8_variance16x16_armv6(second_pass, 16, dst_ptr,
-                                       dst_pixels_per_line, sse);
+        var = vpx_variance16x16_media(second_pass, 16, dst_ptr,
+                                      dst_pixels_per_line, sse);
     }
     return var;
 }
 
-#endif /* HAVE_MEDIA */
+#endif  // HAVE_MEDIA
 
 
 #if HAVE_NEON
@@ -129,4 +133,5 @@ unsigned int vp8_sub_pixel_variance16x16_neon
     return vp8_sub_pixel_variance16x16_neon_func(src_ptr, src_pixels_per_line, xoffset, yoffset, dst_ptr, dst_pixels_per_line, sse);
 }
 
-#endif
+#endif  // HAVE_NEON
+#endif  // CONFIG_VP8_ENCODER
diff --git a/vp8/common/mfqe.c b/vp8/common/mfqe.c
index d12dea193..5c0680f42 100644
--- a/vp8/common/mfqe.c
+++ b/vp8/common/mfqe.c
@@ -151,14 +151,14 @@ static void multiframe_quality_enhance_block
 
     if (blksize == 16)
     {
-        actd = (vp8_variance16x16(yd, yd_stride, VP8_ZEROS, 0, &sse)+128)>>8;
-        act = (vp8_variance16x16(y, y_stride, VP8_ZEROS, 0, &sse)+128)>>8;
+        actd = (vpx_variance16x16(yd, yd_stride, VP8_ZEROS, 0, &sse)+128)>>8;
+        act = (vpx_variance16x16(y, y_stride, VP8_ZEROS, 0, &sse)+128)>>8;
 #ifdef USE_SSD
-        vp8_variance16x16(y, y_stride, yd, yd_stride, &sse);
+        vpx_variance16x16(y, y_stride, yd, yd_stride, &sse);
         sad = (sse + 128)>>8;
-        vp8_variance8x8(u, uv_stride, ud, uvd_stride, &sse);
+        vpx_variance8x8(u, uv_stride, ud, uvd_stride, &sse);
         usad = (sse + 32)>>6;
-        vp8_variance8x8(v, uv_stride, vd, uvd_stride, &sse);
+        vpx_variance8x8(v, uv_stride, vd, uvd_stride, &sse);
         vsad = (sse + 32)>>6;
 #else
         sad = (vpx_sad16x16(y, y_stride, yd, yd_stride) + 128) >> 8;
@@ -168,14 +168,14 @@ static void multiframe_quality_enhance_block
     }
     else /* if (blksize == 8) */
     {
-        actd = (vp8_variance8x8(yd, yd_stride, VP8_ZEROS, 0, &sse)+32)>>6;
-        act = (vp8_variance8x8(y, y_stride, VP8_ZEROS, 0, &sse)+32)>>6;
+        actd = (vpx_variance8x8(yd, yd_stride, VP8_ZEROS, 0, &sse)+32)>>6;
+        act = (vpx_variance8x8(y, y_stride, VP8_ZEROS, 0, &sse)+32)>>6;
 #ifdef USE_SSD
-        vp8_variance8x8(y, y_stride, yd, yd_stride, &sse);
+        vpx_variance8x8(y, y_stride, yd, yd_stride, &sse);
         sad = (sse + 32)>>6;
-        vp8_variance4x4(u, uv_stride, ud, uvd_stride, &sse);
+        vpx_variance4x4(u, uv_stride, ud, uvd_stride, &sse);
         usad = (sse + 8)>>4;
-        vp8_variance4x4(v, uv_stride, vd, uvd_stride, &sse);
+        vpx_variance4x4(v, uv_stride, vd, uvd_stride, &sse);
         vsad = (sse + 8)>>4;
 #else
         sad = (vpx_sad8x8(y, y_stride, yd, yd_stride) + 32) >> 6;
diff --git a/vp8/common/rtcd_defs.pl b/vp8/common/rtcd_defs.pl
index c9f14d58a..4b820338e 100644
--- a/vp8/common/rtcd_defs.pl
+++ b/vp8/common/rtcd_defs.pl
@@ -237,31 +237,6 @@ specialize qw/vp8_bilinear_predict4x4 mmx media neon/;
 $vp8_bilinear_predict4x4_media=vp8_bilinear_predict4x4_armv6;
 
 #
-# Whole-pixel Variance
-#
-add_proto qw/unsigned int vp8_variance4x4/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int  ref_stride, unsigned int *sse";
-specialize qw/vp8_variance4x4 mmx sse2/;
-$vp8_variance4x4_sse2=vp8_variance4x4_wmt;
-
-add_proto qw/unsigned int vp8_variance8x8/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int  ref_stride, unsigned int *sse";
-specialize qw/vp8_variance8x8 mmx sse2 media neon/;
-$vp8_variance8x8_sse2=vp8_variance8x8_wmt;
-$vp8_variance8x8_media=vp8_variance8x8_armv6;
-
-add_proto qw/unsigned int vp8_variance8x16/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int  ref_stride, unsigned int *sse";
-specialize qw/vp8_variance8x16 mmx sse2 neon/;
-$vp8_variance8x16_sse2=vp8_variance8x16_wmt;
-
-add_proto qw/unsigned int vp8_variance16x8/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int  ref_stride, unsigned int *sse";
-specialize qw/vp8_variance16x8 mmx sse2 neon/;
-$vp8_variance16x8_sse2=vp8_variance16x8_wmt;
-
-add_proto qw/unsigned int vp8_variance16x16/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int  ref_stride, unsigned int *sse";
-specialize qw/vp8_variance16x16 mmx sse2 media neon/;
-$vp8_variance16x16_sse2=vp8_variance16x16_wmt;
-$vp8_variance16x16_media=vp8_variance16x16_armv6;
-
-#
 # Sub-pixel Variance
 #
 add_proto qw/unsigned int vp8_sub_pixel_variance4x4/, "const unsigned char  *src_ptr, int  source_stride, int  xoffset, int  yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse";
@@ -309,26 +284,12 @@ $vp8_variance_halfpixvar16x16_hv_media=vp8_variance_halfpixvar16x16_hv_armv6;
 if (vpx_config("CONFIG_VP8_ENCODER") eq "yes") {
 
 #
-# Sum of squares (vector)
-#
-add_proto qw/unsigned int vp8_get_mb_ss/, "const short *";
-specialize qw/vp8_get_mb_ss mmx sse2/;
-
-#
 # SSE (Sum Squared Error)
 #
 add_proto qw/unsigned int vp8_sub_pixel_mse16x16/, "const unsigned char  *src_ptr, int  source_stride, int  xoffset, int  yoffset, const unsigned char *ref_ptr, int Refstride, unsigned int *sse";
 specialize qw/vp8_sub_pixel_mse16x16 mmx sse2/;
 $vp8_sub_pixel_mse16x16_sse2=vp8_sub_pixel_mse16x16_wmt;
 
-add_proto qw/unsigned int vp8_mse16x16/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int  ref_stride, unsigned int *sse";
-specialize qw/vp8_mse16x16 mmx sse2 media neon/;
-$vp8_mse16x16_sse2=vp8_mse16x16_wmt;
-$vp8_mse16x16_media=vp8_mse16x16_armv6;
-
-add_proto qw/unsigned int vp8_get4x4sse_cs/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int  ref_stride";
-specialize qw/vp8_get4x4sse_cs mmx neon/;
-
 #
 # Block copy
 #
diff --git a/vp8/common/variance.h b/vp8/common/variance.h
index b62cc6136..c6c9f41bf 100644
--- a/vp8/common/variance.h
+++ b/vp8/common/variance.h
@@ -39,6 +39,7 @@ typedef void (*vpx_sad_multi_fn_t)(
     const unsigned char *ref_array,
     int  ref_stride,
     unsigned int *sad_array);
+
 typedef void (*vpx_sad_multi_d_fn_t)
     (
      const unsigned char *src_ptr,
@@ -48,7 +49,7 @@ typedef void (*vpx_sad_multi_d_fn_t)
      unsigned int *sad_array
     );
 
-typedef unsigned int (*vp8_variance_fn_t)
+typedef unsigned int (*vpx_variance_fn_t)
     (
      const unsigned char *src_ptr,
      int source_stride,
@@ -68,37 +69,14 @@ typedef unsigned int (*vp8_subpixvariance_fn_t)
       unsigned int *sse
     );
 
-typedef void (*vp8_ssimpf_fn_t)
-      (
-        unsigned char *s,
-        int sp,
-        unsigned char *r,
-        int rp,
-        unsigned long *sum_s,
-        unsigned long *sum_r,
-        unsigned long *sum_sq_s,
-        unsigned long *sum_sq_r,
-        unsigned long *sum_sxr
-      );
-
-typedef unsigned int (*vp8_getmbss_fn_t)(const short *);
-
-typedef unsigned int (*vp8_get16x16prederror_fn_t)
-    (
-     const unsigned char *src_ptr,
-     int source_stride,
-     const unsigned char *ref_ptr,
-     int  ref_stride
-    );
-
 typedef struct variance_vtable
 {
     vpx_sad_fn_t            sdf;
-    vp8_variance_fn_t       vf;
+    vpx_variance_fn_t       vf;
     vp8_subpixvariance_fn_t svf;
-    vp8_variance_fn_t       svf_halfpix_h;
-    vp8_variance_fn_t       svf_halfpix_v;
-    vp8_variance_fn_t       svf_halfpix_hv;
+    vpx_variance_fn_t       svf_halfpix_h;
+    vpx_variance_fn_t       svf_halfpix_v;
+    vpx_variance_fn_t       svf_halfpix_hv;
     vpx_sad_multi_fn_t      sdx3f;
     vpx_sad_multi_fn_t      sdx8f;
     vpx_sad_multi_d_fn_t    sdx4df;
diff --git a/vp8/common/variance_c.c b/vp8/common/variance_c.c
index dc95bfeb3..79d1ca00c 100644
--- a/vp8/common/variance_c.c
+++ b/vp8/common/variance_c.c
@@ -8,44 +8,34 @@
  *  be found in the AUTHORS file in the root of the source tree.
  */
 
-
 #include "./vp8_rtcd.h"
 #include "filter.h"
 #include "variance.h"
 
-
-unsigned int vp8_get_mb_ss_c
-(
-    const short *src_ptr
-)
-{
-    unsigned int i = 0, sum = 0;
-
-    do
-    {
-        sum += (src_ptr[i] * src_ptr[i]);
-        i++;
-    }
-    while (i < 256);
-
-    return sum;
+/* This is a bad idea.
+ * ctz = count trailing zeros */
+static int ctz(int a) {
+  int b = 0;
+  while (a != 1) {
+    a >>= 1;
+    b++;
+  }
+  return b;
 }
 
-
-static void variance(
+static unsigned int variance(
     const unsigned char *src_ptr,
     int  source_stride,
     const unsigned char *ref_ptr,
     int  recon_stride,
     int  w,
     int  h,
-    unsigned int *sse,
-    int *sum)
+    unsigned int *sse)
 {
     int i, j;
-    int diff;
+    int diff, sum;
 
-    *sum = 0;
+    sum = 0;
     *sse = 0;
 
     for (i = 0; i < h; i++)
@@ -53,114 +43,17 @@ static void variance(
         for (j = 0; j < w; j++)
         {
             diff = src_ptr[j] - ref_ptr[j];
-            *sum += diff;
+            sum += diff;
             *sse += diff * diff;
         }
 
         src_ptr += source_stride;
         ref_ptr += recon_stride;
     }
-}
-
 
-unsigned int vp8_variance16x16_c(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int var;
-    int avg;
-
-
-    variance(src_ptr, source_stride, ref_ptr, recon_stride, 16, 16, &var, &avg);
-    *sse = var;
-    return (var - (((unsigned int)avg * avg) >> 8));
-}
-
-unsigned int vp8_variance8x16_c(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int var;
-    int avg;
-
-
-    variance(src_ptr, source_stride, ref_ptr, recon_stride, 8, 16, &var, &avg);
-    *sse = var;
-    return (var - (((unsigned int)avg * avg) >> 7));
+    return (*sse - (((unsigned int)sum * sum) >> (int)((ctz(w) + ctz(h)))));
 }
 
-unsigned int vp8_variance16x8_c(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int var;
-    int avg;
-
-
-    variance(src_ptr, source_stride, ref_ptr, recon_stride, 16, 8, &var, &avg);
-    *sse = var;
-    return (var - (((unsigned int)avg * avg) >> 7));
-}
-
-
-unsigned int vp8_variance8x8_c(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int var;
-    int avg;
-
-
-    variance(src_ptr, source_stride, ref_ptr, recon_stride, 8, 8, &var, &avg);
-    *sse = var;
-    return (var - (((unsigned int)avg * avg) >> 6));
-}
-
-unsigned int vp8_variance4x4_c(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int var;
-    int avg;
-
-
-    variance(src_ptr, source_stride, ref_ptr, recon_stride, 4, 4, &var, &avg);
-    *sse = var;
-    return (var - (((unsigned int)avg * avg) >> 4));
-}
-
-
-unsigned int vp8_mse16x16_c(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int var;
-    int avg;
-
-    variance(src_ptr, source_stride, ref_ptr, recon_stride, 16, 16, &var, &avg);
-    *sse = var;
-    return var;
-}
-
-
 /****************************************************************************
  *
  *  ROUTINE       : filter_block2d_bil_first_pass
@@ -304,7 +197,7 @@ unsigned int vp8_sub_pixel_variance4x4_c
     /* Now filter Verticaly */
     var_filter_block2d_bil_second_pass(FData3, temp2, 4,  4,  4,  4, VFilter);
 
-    return vp8_variance4x4_c(temp2, 4, dst_ptr, dst_pixels_per_line, sse);
+    return variance(temp2, 4, dst_ptr, dst_pixels_per_line, 4, 4, sse);
 }
 
 
@@ -329,7 +222,7 @@ unsigned int vp8_sub_pixel_variance8x8_c
     var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 9, 8, HFilter);
     var_filter_block2d_bil_second_pass(FData3, temp2, 8, 8, 8, 8, VFilter);
 
-    return vp8_variance8x8_c(temp2, 8, dst_ptr, dst_pixels_per_line, sse);
+    return variance(temp2, 8, dst_ptr, dst_pixels_per_line, 8, 8, sse);
 }
 
 unsigned int vp8_sub_pixel_variance16x16_c
@@ -353,7 +246,7 @@ unsigned int vp8_sub_pixel_variance16x16_c
     var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 17, 16, HFilter);
     var_filter_block2d_bil_second_pass(FData3, temp2, 16, 16, 16, 16, VFilter);
 
-    return vp8_variance16x16_c(temp2, 16, dst_ptr, dst_pixels_per_line, sse);
+    return variance(temp2, 16, dst_ptr, dst_pixels_per_line, 16, 16, sse);
 }
 
 
@@ -429,7 +322,7 @@ unsigned int vp8_sub_pixel_variance16x8_c
     var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 9, 16, HFilter);
     var_filter_block2d_bil_second_pass(FData3, temp2, 16, 16, 8, 16, VFilter);
 
-    return vp8_variance16x8_c(temp2, 16, dst_ptr, dst_pixels_per_line, sse);
+    return variance(temp2, 16, dst_ptr, dst_pixels_per_line, 16, 8, sse);
 }
 
 unsigned int vp8_sub_pixel_variance8x16_c
@@ -455,5 +348,5 @@ unsigned int vp8_sub_pixel_variance8x16_c
     var_filter_block2d_bil_first_pass(src_ptr, FData3, src_pixels_per_line, 1, 17, 8, HFilter);
     var_filter_block2d_bil_second_pass(FData3, temp2, 8, 8, 16, 8, VFilter);
 
-    return vp8_variance8x16_c(temp2, 8, dst_ptr, dst_pixels_per_line, sse);
+    return variance(temp2, 8, dst_ptr, dst_pixels_per_line, 8, 16, sse);
 }
diff --git a/vp8/common/x86/variance_impl_mmx.asm b/vp8/common/x86/variance_impl_mmx.asm
index 7d5e6810b..97f25275d 100644
--- a/vp8/common/x86/variance_impl_mmx.asm
+++ b/vp8/common/x86/variance_impl_mmx.asm
@@ -11,504 +11,6 @@
 
 %include "vpx_ports/x86_abi_support.asm"
 
-;unsigned int vp8_get_mb_ss_mmx( short *src_ptr )
-global sym(vp8_get_mb_ss_mmx) PRIVATE
-sym(vp8_get_mb_ss_mmx):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 7
-    GET_GOT     rbx
-    push rsi
-    push rdi
-    sub         rsp, 8
-    ; end prolog
-
-        mov         rax, arg(0) ;src_ptr
-        mov         rcx, 16
-        pxor        mm4, mm4
-
-.NEXTROW:
-        movq        mm0, [rax]
-        movq        mm1, [rax+8]
-        movq        mm2, [rax+16]
-        movq        mm3, [rax+24]
-        pmaddwd     mm0, mm0
-        pmaddwd     mm1, mm1
-        pmaddwd     mm2, mm2
-        pmaddwd     mm3, mm3
-
-        paddd       mm4, mm0
-        paddd       mm4, mm1
-        paddd       mm4, mm2
-        paddd       mm4, mm3
-
-        add         rax, 32
-        dec         rcx
-        ja          .NEXTROW
-        movq        QWORD PTR [rsp], mm4
-
-        ;return sum[0]+sum[1];
-        movsxd      rax, dword ptr [rsp]
-        movsxd      rcx, dword ptr [rsp+4]
-        add         rax, rcx
-
-
-    ; begin epilog
-    add rsp, 8
-    pop rdi
-    pop rsi
-    RESTORE_GOT
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
-
-;unsigned int vp8_get8x8var_mmx
-;(
-;    unsigned char *src_ptr,
-;    int  source_stride,
-;    unsigned char *ref_ptr,
-;    int  recon_stride,
-;    unsigned int *SSE,
-;    int *Sum
-;)
-global sym(vp8_get8x8var_mmx) PRIVATE
-sym(vp8_get8x8var_mmx):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 6
-    push rsi
-    push rdi
-    push rbx
-    sub         rsp, 16
-    ; end prolog
-
-
-        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
-        movq        mm2, mm0                    ; Take copies
-        movq        mm3, mm1                    ; Take copies
-
-        punpcklbw   mm0, mm6                    ; unpack to higher prrcision
-        punpcklbw   mm1, mm6
-        punpckhbw   mm2, mm6                    ; unpack to higher prrcision
-        punpckhbw   mm3, mm6
-        psubsw      mm0, mm1                    ; A-B (low order) to MM0
-        psubsw      mm2, mm3                    ; A-B (high order) to MM2
-
-        paddw       mm5, mm0                    ; accumulate differences in mm5
-        paddw       mm5, mm2                    ; accumulate differences in mm5
-
-        pmaddwd     mm0, mm0                    ; square and accumulate
-        pmaddwd     mm2, mm2                    ; 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
-        paddd       mm7, mm0                    ; accumulate in mm7
-        paddd       mm7, mm2                    ; accumulate in mm7
-
-
-        ; Row 2
-        movq        mm0, [rax]                  ; Copy eight bytes to mm0
-        movq        mm2, mm0                    ; Take copies
-        movq        mm3, mm1                    ; Take copies
-
-        punpcklbw   mm0, mm6                    ; unpack to higher prrcision
-        punpcklbw   mm1, mm6
-        punpckhbw   mm2, mm6                    ; unpack to higher prrcision
-        punpckhbw   mm3, mm6
-        psubsw      mm0, mm1                    ; A-B (low order) to MM0
-        psubsw      mm2, mm3                    ; A-B (high order) to MM2
-
-        paddw       mm5, mm0                    ; accumulate differences in mm5
-        paddw       mm5, mm2                    ; accumulate differences in mm5
-
-        pmaddwd     mm0, mm0                    ; square and accumulate
-        pmaddwd     mm2, mm2                    ; 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
-        paddd       mm7, mm0                    ; accumulate in mm7
-        paddd       mm7, mm2                    ; accumulate in mm7
-
-        ; Row 3
-        movq        mm0, [rax]                  ; Copy eight bytes to mm0
-        movq        mm2, mm0                    ; Take copies
-        movq        mm3, mm1                    ; Take copies
-
-        punpcklbw   mm0, mm6                    ; unpack to higher prrcision
-        punpcklbw   mm1, mm6
-        punpckhbw   mm2, mm6                    ; unpack to higher prrcision
-        punpckhbw   mm3, mm6
-        psubsw      mm0, mm1                    ; A-B (low order) to MM0
-        psubsw      mm2, mm3                    ; A-B (high order) to MM2
-
-        paddw       mm5, mm0                    ; accumulate differences in mm5
-        paddw       mm5, mm2                    ; accumulate differences in mm5
-
-        pmaddwd     mm0, mm0                    ; square and accumulate
-        pmaddwd     mm2, mm2                    ; 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
-        paddd       mm7, mm0                    ; accumulate in mm7
-        paddd       mm7, mm2                    ; accumulate in mm7
-
-        ; Row 4
-        movq        mm0, [rax]                  ; Copy eight bytes to mm0
-        movq        mm2, mm0                    ; Take copies
-        movq        mm3, mm1                    ; Take copies
-
-        punpcklbw   mm0, mm6                    ; unpack to higher prrcision
-        punpcklbw   mm1, mm6
-        punpckhbw   mm2, mm6                    ; unpack to higher prrcision
-        punpckhbw   mm3, mm6
-        psubsw      mm0, mm1                    ; A-B (low order) to MM0
-        psubsw      mm2, mm3                    ; A-B (high order) to MM2
-
-        paddw       mm5, mm0                    ; accumulate differences in mm5
-        paddw       mm5, mm2                    ; accumulate differences in mm5
-
-        pmaddwd     mm0, mm0                    ; square and accumulate
-        pmaddwd     mm2, mm2                    ; 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
-        paddd       mm7, mm0                    ; accumulate in mm7
-        paddd       mm7, mm2                    ; accumulate in mm7
-
-        ; Row 5
-        movq        mm0, [rax]                  ; Copy eight bytes to mm0
-        movq        mm2, mm0                    ; Take copies
-        movq        mm3, mm1                    ; Take copies
-
-        punpcklbw   mm0, mm6                    ; unpack to higher prrcision
-        punpcklbw   mm1, mm6
-        punpckhbw   mm2, mm6                    ; unpack to higher prrcision
-        punpckhbw   mm3, mm6
-        psubsw      mm0, mm1                    ; A-B (low order) to MM0
-        psubsw      mm2, mm3                    ; A-B (high order) to MM2
-
-        paddw       mm5, mm0                    ; accumulate differences in mm5
-        paddw       mm5, mm2                    ; accumulate differences in mm5
-
-        pmaddwd     mm0, mm0                    ; square and accumulate
-        pmaddwd     mm2, mm2                    ; 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
-        ;              movq        mm4, [rbx + rdx]
-        paddd       mm7, mm0                    ; accumulate in mm7
-        paddd       mm7, mm2                    ; accumulate in mm7
-
-        ; Row 6
-        movq        mm0, [rax]                  ; Copy eight bytes to mm0
-        movq        mm2, mm0                    ; Take copies
-        movq        mm3, mm1                    ; Take copies
-
-        punpcklbw   mm0, mm6                    ; unpack to higher prrcision
-        punpcklbw   mm1, mm6
-        punpckhbw   mm2, mm6                    ; unpack to higher prrcision
-        punpckhbw   mm3, mm6
-        psubsw      mm0, mm1                    ; A-B (low order) to MM0
-        psubsw      mm2, mm3                    ; A-B (high order) to MM2
-
-        paddw       mm5, mm0                    ; accumulate differences in mm5
-        paddw       mm5, mm2                    ; accumulate differences in mm5
-
-        pmaddwd     mm0, mm0                    ; square and accumulate
-        pmaddwd     mm2, mm2                    ; 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
-        paddd       mm7, mm0                    ; accumulate in mm7
-        paddd       mm7, mm2                    ; accumulate in mm7
-
-        ; Row 7
-        movq        mm0, [rax]                  ; Copy eight bytes to mm0
-        movq        mm2, mm0                    ; Take copies
-        movq        mm3, mm1                    ; Take copies
-
-        punpcklbw   mm0, mm6                    ; unpack to higher prrcision
-        punpcklbw   mm1, mm6
-        punpckhbw   mm2, mm6                    ; unpack to higher prrcision
-        punpckhbw   mm3, mm6
-        psubsw      mm0, mm1                    ; A-B (low order) to MM0
-        psubsw      mm2, mm3                    ; A-B (high order) to MM2
-
-        paddw       mm5, mm0                    ; accumulate differences in mm5
-        paddw       mm5, mm2                    ; accumulate differences in mm5
-
-        pmaddwd     mm0, mm0                    ; square and accumulate
-        pmaddwd     mm2, mm2                    ; 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
-        paddd       mm7, mm0                    ; accumulate in mm7
-        paddd       mm7, mm2                    ; accumulate in mm7
-
-        ; Row 8
-        movq        mm0, [rax]                  ; Copy eight bytes to mm0
-        movq        mm2, mm0                    ; Take copies
-        movq        mm3, mm1                    ; Take copies
-
-        punpcklbw   mm0, mm6                    ; unpack to higher prrcision
-        punpcklbw   mm1, mm6
-        punpckhbw   mm2, mm6                    ; unpack to higher prrcision
-        punpckhbw   mm3, mm6
-        psubsw      mm0, mm1                    ; A-B (low order) to MM0
-        psubsw      mm2, mm3                    ; A-B (high order) to MM2
-
-        paddw       mm5, mm0                    ; accumulate differences in mm5
-        paddw       mm5, mm2                    ; accumulate differences in mm5
-
-        pmaddwd     mm0, mm0                    ; square and accumulate
-        pmaddwd     mm2, mm2                    ; square and accumulate
-        add         rbx,rdx                     ; Inc pointer into ref data
-        add         rax,rcx                     ; Inc pointer into the new data
-        paddd       mm7, mm0                    ; accumulate in mm7
-        paddd       mm7, mm2                    ; accumulate in mm7
-
-        ; Now accumulate the final results.
-        movq        QWORD PTR [rsp+8], mm5      ; copy back accumulated results into normal memory
-        movq        QWORD PTR [rsp], mm7        ; copy back accumulated results into normal memory
-        movsx       rdx, WORD PTR [rsp+8]
-        movsx       rcx, WORD PTR [rsp+10]
-        movsx       rbx, WORD PTR [rsp+12]
-        movsx       rax, WORD PTR [rsp+14]
-        add         rdx, rcx
-        add         rbx, rax
-        add         rdx, rbx    ;XSum
-        movsxd      rax, DWORD PTR [rsp]
-        movsxd      rcx, DWORD PTR [rsp+4]
-        add         rax, rcx    ;XXSum
-        mov         rsi, arg(4) ;SSE
-        mov         rdi, arg(5) ;Sum
-        mov         dword ptr [rsi], eax
-        mov         dword ptr [rdi], edx
-        xor         rax, rax    ; return 0
-
-
-    ; begin epilog
-    add rsp, 16
-    pop rbx
-    pop rdi
-    pop rsi
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
-
-
-;unsigned int
-;vp8_get4x4var_mmx
-;(
-;    unsigned char *src_ptr,
-;    int  source_stride,
-;    unsigned char *ref_ptr,
-;    int  recon_stride,
-;    unsigned int *SSE,
-;    int *Sum
-;)
-global sym(vp8_get4x4var_mmx) PRIVATE
-sym(vp8_get4x4var_mmx):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 6
-    push rsi
-    push rdi
-    push rbx
-    sub         rsp, 16
-    ; end prolog
-
-
-        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
-        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
-        movd        mm1, [rbx]                  ; Copy four bytes to mm1
-        paddd       mm7, mm0                    ; accumulate in mm7
-
-
-        ; Row 2
-        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
-        movd        mm1, [rbx]                  ; Copy four bytes to mm1
-        paddd       mm7, mm0                    ; accumulate in mm7
-
-        ; Row 3
-        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
-        movd        mm1, [rbx]                  ; Copy four bytes to mm1
-        paddd       mm7, mm0                    ; accumulate in mm7
-
-        ; Row 4
-        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
-
-
-        ; Now accumulate the final results.
-        movq        QWORD PTR [rsp+8], mm5      ; copy back accumulated results into normal memory
-        movq        QWORD PTR [rsp], mm7        ; copy back accumulated results into normal memory
-        movsx       rdx, WORD PTR [rsp+8]
-        movsx       rcx, WORD PTR [rsp+10]
-        movsx       rbx, WORD PTR [rsp+12]
-        movsx       rax, WORD PTR [rsp+14]
-        add         rdx, rcx
-        add         rbx, rax
-        add         rdx, rbx    ;XSum
-        movsxd      rax, DWORD PTR [rsp]
-        movsxd      rcx, DWORD PTR [rsp+4]
-        add         rax, rcx    ;XXSum
-        mov         rsi, arg(4) ;SSE
-        mov         rdi, arg(5) ;Sum
-        mov         dword ptr [rsi], eax
-        mov         dword ptr [rdi], edx
-        xor         rax, rax    ; return 0
-
-
-    ; begin epilog
-    add rsp, 16
-    pop rbx
-    pop rdi
-    pop rsi
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
-
-
-;unsigned int
-;vp8_get4x4sse_cs_mmx
-;(
-;    unsigned char *src_ptr,
-;    int  source_stride,
-;    unsigned char *ref_ptr,
-;    int  recon_stride
-;)
-global sym(vp8_get4x4sse_cs_mmx) PRIVATE
-sym(vp8_get4x4sse_cs_mmx):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 4
-    push rsi
-    push rdi
-    push rbx
-    ; end prolog
-
-
-        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
-        movd        mm0, [rax]                  ; Copy eight bytes to mm0
-        movd        mm1, [rbx]                  ; Copy eight bytes to mm1
-        punpcklbw   mm0, mm6                    ; unpack to higher prrcision
-        punpcklbw   mm1, mm6
-        psubsw      mm0, mm1                    ; A-B (low order) to MM0
-        pmaddwd     mm0, mm0                    ; square and accumulate
-        add         rbx,rdx                     ; Inc pointer into ref data
-        add         rax,rcx                     ; Inc pointer into the new data
-        movd        mm1, [rbx]                  ; Copy eight bytes to mm1
-        paddd       mm7, mm0                    ; accumulate in mm7
-
-        ; Row 2
-        movd        mm0, [rax]                  ; Copy eight bytes to mm0
-        punpcklbw   mm0, mm6                    ; unpack to higher prrcision
-        punpcklbw   mm1, mm6
-        psubsw      mm0, mm1                    ; A-B (low order) to MM0
-        pmaddwd     mm0, mm0                    ; square and accumulate
-        add         rbx,rdx                     ; Inc pointer into ref data
-        add         rax,rcx                     ; Inc pointer into the new data
-        movd        mm1, [rbx]                  ; Copy eight bytes to mm1
-        paddd       mm7, mm0                    ; accumulate in mm7
-
-        ; Row 3
-        movd        mm0, [rax]                  ; Copy eight bytes to mm0
-        punpcklbw   mm1, mm6
-        punpcklbw   mm0, mm6                    ; unpack to higher prrcision
-        psubsw      mm0, mm1                    ; A-B (low order) to MM0
-
-        pmaddwd     mm0, mm0                    ; square and accumulate
-        add         rbx,rdx                     ; Inc pointer into ref data
-        add         rax,rcx                     ; Inc pointer into the new data
-        movd        mm1, [rbx]                  ; Copy eight bytes to mm1
-        paddd       mm7, mm0                    ; accumulate in mm7
-
-        ; Row 4
-        movd        mm0, [rax]                  ; Copy eight bytes to mm0
-        punpcklbw   mm0, mm6                    ; unpack to higher prrcision
-        punpcklbw   mm1, mm6
-        psubsw      mm0, mm1                    ; A-B (low order) to MM0
-        pmaddwd     mm0, mm0                    ; square and accumulate
-        paddd       mm7, mm0                    ; accumulate in mm7
-
-        movq        mm0,    mm7                 ;
-        psrlq       mm7,    32
-
-        paddd       mm0,    mm7
-        movq        rax,    mm0
-
-
-    ; begin epilog
-    pop rbx
-    pop rdi
-    pop rsi
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
 %define mmx_filter_shift            7
 
 ;void vp8_filter_block2d_bil4x4_var_mmx
diff --git a/vp8/common/x86/variance_impl_sse2.asm b/vp8/common/x86/variance_impl_sse2.asm
index 761433c11..26de5e860 100644
--- a/vp8/common/x86/variance_impl_sse2.asm
+++ b/vp8/common/x86/variance_impl_sse2.asm
@@ -13,393 +13,6 @@
 
 %define xmm_filter_shift            7
 
-;unsigned int vp8_get_mb_ss_sse2
-;(
-;    short *src_ptr
-;)
-global sym(vp8_get_mb_ss_sse2) PRIVATE
-sym(vp8_get_mb_ss_sse2):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 1
-    GET_GOT     rbx
-    push rsi
-    push rdi
-    sub         rsp, 16
-    ; end prolog
-
-
-        mov         rax, arg(0) ;[src_ptr]
-        mov         rcx, 8
-        pxor        xmm4, xmm4
-
-.NEXTROW:
-        movdqa      xmm0, [rax]
-        movdqa      xmm1, [rax+16]
-        movdqa      xmm2, [rax+32]
-        movdqa      xmm3, [rax+48]
-        pmaddwd     xmm0, xmm0
-        pmaddwd     xmm1, xmm1
-        pmaddwd     xmm2, xmm2
-        pmaddwd     xmm3, xmm3
-
-        paddd       xmm0, xmm1
-        paddd       xmm2, xmm3
-        paddd       xmm4, xmm0
-        paddd       xmm4, xmm2
-
-        add         rax, 0x40
-        dec         rcx
-        ja          .NEXTROW
-
-        movdqa      xmm3,xmm4
-        psrldq      xmm4,8
-        paddd       xmm4,xmm3
-        movdqa      xmm3,xmm4
-        psrldq      xmm4,4
-        paddd       xmm4,xmm3
-        movq        rax,xmm4
-
-
-    ; begin epilog
-    add rsp, 16
-    pop rdi
-    pop rsi
-    RESTORE_GOT
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
-
-;unsigned int vp8_get16x16var_sse2
-;(
-;    unsigned char   *  src_ptr,
-;    int             source_stride,
-;    unsigned char   *  ref_ptr,
-;    int             recon_stride,
-;    unsigned int    *  SSE,
-;    int             *  Sum
-;)
-global sym(vp8_get16x16var_sse2) PRIVATE
-sym(vp8_get16x16var_sse2):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 6
-    SAVE_XMM 7
-    push rbx
-    push rsi
-    push rdi
-    ; end prolog
-
-        mov         rsi,            arg(0) ;[src_ptr]
-        mov         rdi,            arg(2) ;[ref_ptr]
-
-        movsxd      rax,            DWORD PTR arg(1) ;[source_stride]
-        movsxd      rdx,            DWORD PTR arg(3) ;[recon_stride]
-
-        ; Prefetch data
-        lea             rcx,    [rax+rax*2]
-        prefetcht0      [rsi]
-        prefetcht0      [rsi+rax]
-        prefetcht0      [rsi+rax*2]
-        prefetcht0      [rsi+rcx]
-        lea             rbx,    [rsi+rax*4]
-        prefetcht0      [rbx]
-        prefetcht0      [rbx+rax]
-        prefetcht0      [rbx+rax*2]
-        prefetcht0      [rbx+rcx]
-
-        lea             rcx,    [rdx+rdx*2]
-        prefetcht0      [rdi]
-        prefetcht0      [rdi+rdx]
-        prefetcht0      [rdi+rdx*2]
-        prefetcht0      [rdi+rcx]
-        lea             rbx,    [rdi+rdx*4]
-        prefetcht0      [rbx]
-        prefetcht0      [rbx+rdx]
-        prefetcht0      [rbx+rdx*2]
-        prefetcht0      [rbx+rcx]
-
-        pxor        xmm0,           xmm0                        ; clear xmm0 for unpack
-        pxor        xmm7,           xmm7                        ; clear xmm7 for accumulating diffs
-
-        pxor        xmm6,           xmm6                        ; clear xmm6 for accumulating sse
-        mov         rcx,            16
-
-.var16loop:
-        movdqu      xmm1,           XMMWORD PTR [rsi]
-        movdqu      xmm2,           XMMWORD PTR [rdi]
-
-        prefetcht0      [rsi+rax*8]
-        prefetcht0      [rdi+rdx*8]
-
-        movdqa      xmm3,           xmm1
-        movdqa      xmm4,           xmm2
-
-
-        punpcklbw   xmm1,           xmm0
-        punpckhbw   xmm3,           xmm0
-
-        punpcklbw   xmm2,           xmm0
-        punpckhbw   xmm4,           xmm0
-
-
-        psubw       xmm1,           xmm2
-        psubw       xmm3,           xmm4
-
-        paddw       xmm7,           xmm1
-        pmaddwd     xmm1,           xmm1
-
-        paddw       xmm7,           xmm3
-        pmaddwd     xmm3,           xmm3
-
-        paddd       xmm6,           xmm1
-        paddd       xmm6,           xmm3
-
-        add         rsi,            rax
-        add         rdi,            rdx
-
-        sub         rcx,            1
-        jnz         .var16loop
-
-
-        movdqa      xmm1,           xmm6
-        pxor        xmm6,           xmm6
-
-        pxor        xmm5,           xmm5
-        punpcklwd   xmm6,           xmm7
-
-        punpckhwd   xmm5,           xmm7
-        psrad       xmm5,           16
-
-        psrad       xmm6,           16
-        paddd       xmm6,           xmm5
-
-        movdqa      xmm2,           xmm1
-        punpckldq   xmm1,           xmm0
-
-        punpckhdq   xmm2,           xmm0
-        movdqa      xmm7,           xmm6
-
-        paddd       xmm1,           xmm2
-        punpckldq   xmm6,           xmm0
-
-        punpckhdq   xmm7,           xmm0
-        paddd       xmm6,           xmm7
-
-        movdqa      xmm2,           xmm1
-        movdqa      xmm7,           xmm6
-
-        psrldq      xmm1,           8
-        psrldq      xmm6,           8
-
-        paddd       xmm7,           xmm6
-        paddd       xmm1,           xmm2
-
-        mov         rax,            arg(5) ;[Sum]
-        mov         rdi,            arg(4) ;[SSE]
-
-        movd DWORD PTR [rax],       xmm7
-        movd DWORD PTR [rdi],       xmm1
-
-
-    ; begin epilog
-    pop rdi
-    pop rsi
-    pop rbx
-    RESTORE_XMM
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
-
-
-
-;unsigned int vp8_get8x8var_sse2
-;(
-;    unsigned char   *  src_ptr,
-;    int             source_stride,
-;    unsigned char   *  ref_ptr,
-;    int             recon_stride,
-;    unsigned int    *  SSE,
-;    int             *  Sum
-;)
-global sym(vp8_get8x8var_sse2) PRIVATE
-sym(vp8_get8x8var_sse2):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 6
-    SAVE_XMM 7
-    GET_GOT     rbx
-    push rsi
-    push rdi
-    sub         rsp, 16
-    ; end prolog
-
-        mov         rsi,            arg(0) ;[src_ptr]
-        mov         rdi,            arg(2) ;[ref_ptr]
-
-        movsxd      rax,            DWORD PTR arg(1) ;[source_stride]
-        movsxd      rdx,            DWORD PTR arg(3) ;[recon_stride]
-
-        pxor        xmm0,           xmm0                        ; clear xmm0 for unpack
-        pxor        xmm7,           xmm7                        ; clear xmm7 for accumulating diffs
-
-        movq        xmm1,           QWORD PTR [rsi]
-        movq        xmm2,           QWORD PTR [rdi]
-
-        punpcklbw   xmm1,           xmm0
-        punpcklbw   xmm2,           xmm0
-
-        psubsw      xmm1,           xmm2
-        paddw       xmm7,           xmm1
-
-        pmaddwd     xmm1,           xmm1
-
-        movq        xmm2,           QWORD PTR[rsi + rax]
-        movq        xmm3,           QWORD PTR[rdi + rdx]
-
-        punpcklbw   xmm2,           xmm0
-        punpcklbw   xmm3,           xmm0
-
-        psubsw      xmm2,           xmm3
-        paddw       xmm7,           xmm2
-
-        pmaddwd     xmm2,           xmm2
-        paddd       xmm1,           xmm2
-
-
-        movq        xmm2,           QWORD PTR[rsi + rax * 2]
-        movq        xmm3,           QWORD PTR[rdi + rdx * 2]
-
-        punpcklbw   xmm2,           xmm0
-        punpcklbw   xmm3,           xmm0
-
-        psubsw      xmm2,           xmm3
-        paddw       xmm7,           xmm2
-
-        pmaddwd     xmm2,           xmm2
-        paddd       xmm1,           xmm2
-
-
-        lea         rsi,            [rsi + rax * 2]
-        lea         rdi,            [rdi + rdx * 2]
-        movq        xmm2,           QWORD PTR[rsi + rax]
-        movq        xmm3,           QWORD PTR[rdi + rdx]
-
-        punpcklbw   xmm2,           xmm0
-        punpcklbw   xmm3,           xmm0
-
-        psubsw      xmm2,           xmm3
-        paddw       xmm7,           xmm2
-
-        pmaddwd     xmm2,           xmm2
-        paddd       xmm1,           xmm2
-
-        movq        xmm2,           QWORD PTR[rsi + rax *2]
-        movq        xmm3,           QWORD PTR[rdi + rdx *2]
-
-        punpcklbw   xmm2,           xmm0
-        punpcklbw   xmm3,           xmm0
-
-        psubsw      xmm2,           xmm3
-        paddw       xmm7,           xmm2
-
-        pmaddwd     xmm2,           xmm2
-        paddd       xmm1,           xmm2
-
-
-        lea         rsi,            [rsi + rax * 2]
-        lea         rdi,            [rdi + rdx * 2]
-
-
-        movq        xmm2,           QWORD PTR[rsi + rax]
-        movq        xmm3,           QWORD PTR[rdi + rdx]
-
-        punpcklbw   xmm2,           xmm0
-        punpcklbw   xmm3,           xmm0
-
-        psubsw      xmm2,           xmm3
-        paddw       xmm7,           xmm2
-
-        pmaddwd     xmm2,           xmm2
-        paddd       xmm1,           xmm2
-
-        movq        xmm2,           QWORD PTR[rsi + rax *2]
-        movq        xmm3,           QWORD PTR[rdi + rdx *2]
-
-        punpcklbw   xmm2,           xmm0
-        punpcklbw   xmm3,           xmm0
-
-        psubsw      xmm2,           xmm3
-        paddw       xmm7,           xmm2
-
-        pmaddwd     xmm2,           xmm2
-        paddd       xmm1,           xmm2
-
-
-        lea         rsi,            [rsi + rax * 2]
-        lea         rdi,            [rdi + rdx * 2]
-
-        movq        xmm2,           QWORD PTR[rsi + rax]
-        movq        xmm3,           QWORD PTR[rdi + rdx]
-
-        punpcklbw   xmm2,           xmm0
-        punpcklbw   xmm3,           xmm0
-
-        psubsw      xmm2,           xmm3
-        paddw       xmm7,           xmm2
-
-        pmaddwd     xmm2,           xmm2
-        paddd       xmm1,           xmm2
-
-
-        movdqa      xmm6,           xmm7
-        punpcklwd   xmm6,           xmm0
-
-        punpckhwd   xmm7,           xmm0
-        movdqa      xmm2,           xmm1
-
-        paddw       xmm6,           xmm7
-        punpckldq   xmm1,           xmm0
-
-        punpckhdq   xmm2,           xmm0
-        movdqa      xmm7,           xmm6
-
-        paddd       xmm1,           xmm2
-        punpckldq   xmm6,           xmm0
-
-        punpckhdq   xmm7,           xmm0
-        paddw       xmm6,           xmm7
-
-        movdqa      xmm2,           xmm1
-        movdqa      xmm7,           xmm6
-
-        psrldq      xmm1,           8
-        psrldq      xmm6,           8
-
-        paddw       xmm7,           xmm6
-        paddd       xmm1,           xmm2
-
-        mov         rax,            arg(5) ;[Sum]
-        mov         rdi,            arg(4) ;[SSE]
-
-        movq        rdx,            xmm7
-        movsx       rcx,            dx
-
-        mov  dword ptr [rax],       ecx
-        movd DWORD PTR [rdi],       xmm1
-
-    ; begin epilog
-    add rsp, 16
-    pop rdi
-    pop rsi
-    RESTORE_GOT
-    RESTORE_XMM
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
 ;void vp8_filter_block2d_bil_var_sse2
 ;(
 ;    unsigned char *ref_ptr,
diff --git a/vp8/common/x86/variance_mmx.c b/vp8/common/x86/variance_mmx.c
index 10a58b822..25ae5767f 100644
--- a/vp8/common/x86/variance_mmx.c
+++ b/vp8/common/x86/variance_mmx.c
@@ -35,25 +35,6 @@ extern void filter_block1d_v6_mmx
     short *filter
 );
 
-extern unsigned int vp8_get_mb_ss_mmx(const short *src_ptr);
-extern unsigned int vp8_get8x8var_mmx
-(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *SSE,
-    int *Sum
-);
-extern unsigned int vp8_get4x4var_mmx
-(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *SSE,
-    int *Sum
-);
 extern void vp8_filter_block2d_bil4x4_var_mmx
 (
     const unsigned char *ref_ptr,
@@ -78,127 +59,6 @@ extern void vp8_filter_block2d_bil_var_mmx
     unsigned int *sumsquared
 );
 
-
-unsigned int vp8_variance4x4_mmx(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int var;
-    int avg;
-
-    vp8_get4x4var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &var, &avg) ;
-    *sse = var;
-    return (var - (((unsigned int)avg * avg) >> 4));
-
-}
-
-unsigned int vp8_variance8x8_mmx(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int var;
-    int avg;
-
-    vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &var, &avg) ;
-    *sse = var;
-
-    return (var - (((unsigned int)avg * avg) >> 6));
-
-}
-
-unsigned int vp8_mse16x16_mmx(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int sse0, sse1, sse2, sse3, var;
-    int sum0, sum1, sum2, sum3;
-
-
-    vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
-    vp8_get8x8var_mmx(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
-    vp8_get8x8var_mmx(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse2, &sum2) ;
-    vp8_get8x8var_mmx(src_ptr + 8 * source_stride + 8, source_stride, ref_ptr + 8 * recon_stride + 8, recon_stride, &sse3, &sum3);
-
-    var = sse0 + sse1 + sse2 + sse3;
-    *sse = var;
-    return var;
-}
-
-
-unsigned int vp8_variance16x16_mmx(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int sse0, sse1, sse2, sse3, var;
-    int sum0, sum1, sum2, sum3, avg;
-
-
-    vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
-    vp8_get8x8var_mmx(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
-    vp8_get8x8var_mmx(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse2, &sum2) ;
-    vp8_get8x8var_mmx(src_ptr + 8 * source_stride + 8, source_stride, ref_ptr + 8 * recon_stride + 8, recon_stride, &sse3, &sum3);
-
-    var = sse0 + sse1 + sse2 + sse3;
-    avg = sum0 + sum1 + sum2 + sum3;
-    *sse = var;
-    return (var - (((unsigned int)avg * avg) >> 8));
-}
-
-unsigned int vp8_variance16x8_mmx(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int sse0, sse1, var;
-    int sum0, sum1, avg;
-
-    vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
-    vp8_get8x8var_mmx(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
-
-    var = sse0 + sse1;
-    avg = sum0 + sum1;
-    *sse = var;
-    return (var - (((unsigned int)avg * avg) >> 7));
-
-}
-
-
-unsigned int vp8_variance8x16_mmx(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int sse0, sse1, var;
-    int sum0, sum1, avg;
-
-    vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
-    vp8_get8x8var_mmx(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse1, &sum1) ;
-
-    var = sse0 + sse1;
-    avg = sum0 + sum1;
-    *sse = var;
-
-    return (var - (((unsigned int)avg * avg) >> 7));
-
-}
-
-
 unsigned int vp8_sub_pixel_variance4x4_mmx
 (
     const unsigned char  *src_ptr,
diff --git a/vp8/common/x86/variance_sse2.c b/vp8/common/x86/variance_sse2.c
index 6c6539d8e..f6dfb2787 100644
--- a/vp8/common/x86/variance_sse2.c
+++ b/vp8/common/x86/variance_sse2.c
@@ -31,38 +31,6 @@ extern void vp8_filter_block2d_bil4x4_var_mmx
     unsigned int *sumsquared
 );
 
-extern unsigned int vp8_get4x4var_mmx
-(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *SSE,
-    int *Sum
-);
-
-unsigned int vp8_get_mb_ss_sse2
-(
-    const short *src_ptr
-);
-unsigned int vp8_get16x16var_sse2
-(
-    const unsigned char *src_ptr,
-    int source_stride,
-    const unsigned char *ref_ptr,
-    int recon_stride,
-    unsigned int *SSE,
-    int *Sum
-);
-unsigned int vp8_get8x8var_sse2
-(
-    const unsigned char *src_ptr,
-    int source_stride,
-    const unsigned char *ref_ptr,
-    int recon_stride,
-    unsigned int *SSE,
-    int *Sum
-);
 void vp8_filter_block2d_bil_var_sse2
 (
     const unsigned char *ref_ptr,
@@ -136,115 +104,6 @@ void vp8_half_vert_variance16x_h_sse2
     unsigned int *sumsquared
 );
 
-unsigned int vp8_variance4x4_wmt(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int var;
-    int avg;
-
-    vp8_get4x4var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &var, &avg) ;
-    *sse = var;
-    return (var - (((unsigned int)avg * avg) >> 4));
-
-}
-
-unsigned int vp8_variance8x8_wmt
-(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int var;
-    int avg;
-
-    vp8_get8x8var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &var, &avg) ;
-    *sse = var;
-    return (var - (((unsigned int)avg * avg) >> 6));
-
-}
-
-
-unsigned int vp8_variance16x16_wmt
-(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int sse0;
-    int sum0;
-
-
-    vp8_get16x16var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
-    *sse = sse0;
-    return (sse0 - (((unsigned int)sum0 * sum0) >> 8));
-}
-unsigned int vp8_mse16x16_wmt(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-
-    unsigned int sse0;
-    int sum0;
-    vp8_get16x16var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
-    *sse = sse0;
-    return sse0;
-
-}
-
-
-unsigned int vp8_variance16x8_wmt
-(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int sse0, sse1, var;
-    int sum0, sum1, avg;
-
-    vp8_get8x8var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
-    vp8_get8x8var_sse2(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
-
-    var = sse0 + sse1;
-    avg = sum0 + sum1;
-    *sse = var;
-    return (var - (((unsigned int)avg * avg) >> 7));
-
-}
-
-unsigned int vp8_variance8x16_wmt
-(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *sse)
-{
-    unsigned int sse0, sse1, var;
-    int sum0, sum1, avg;
-
-    vp8_get8x8var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
-    vp8_get8x8var_sse2(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse1, &sum1) ;
-
-    var = sse0 + sse1;
-    avg = sum0 + sum1;
-    *sse = var;
-    return (var - (((unsigned int)avg * avg) >> 7));
-
-}
-
 unsigned int vp8_sub_pixel_variance4x4_wmt
 (
     const unsigned char  *src_ptr,
diff --git a/vp8/common/x86/variance_ssse3.c b/vp8/common/x86/variance_ssse3.c
index d8c8da540..2a0df640a 100644
--- a/vp8/common/x86/variance_ssse3.c
+++ b/vp8/common/x86/variance_ssse3.c
@@ -13,15 +13,6 @@
 #include "vp8/common/variance.h"
 #include "vpx_ports/mem.h"
 
-extern unsigned int vp8_get16x16var_sse2
-(
-    const unsigned char *src_ptr,
-    int source_stride,
-    const unsigned char *ref_ptr,
-    int recon_stride,
-    unsigned int *SSE,
-    int *Sum
-);
 extern void vp8_half_horiz_vert_variance16x_h_sse2
 (
     const unsigned char *ref_ptr,
diff --git a/vp8/encoder/arm/armv6/vp8_mse16x16_armv6.asm b/vp8/encoder/arm/armv6/vp8_mse16x16_armv6.asm
deleted file mode 100644
index 000805d4f..000000000
--- a/vp8/encoder/arm/armv6/vp8_mse16x16_armv6.asm
+++ /dev/null
@@ -1,138 +0,0 @@
-;
-;  Copyright (c) 2011 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.
-;
-
-
-    EXPORT  |vp8_mse16x16_armv6|
-
-    ARM
-
-    AREA ||.text||, CODE, READONLY, ALIGN=2
-
-; r0    unsigned char *src_ptr
-; r1    int source_stride
-; r2    unsigned char *ref_ptr
-; r3    int  recon_stride
-; stack unsigned int *sse
-;
-;note: Based on vp8_variance16x16_armv6. In this function, sum is never used.
-;      So, we can remove this part of calculation.
-
-|vp8_mse16x16_armv6| PROC
-
-    push    {r4-r9, lr}
-
-    pld     [r0, r1, lsl #0]
-    pld     [r2, r3, lsl #0]
-
-    mov     r12, #16            ; set loop counter to 16 (=block height)
-    mov     r4, #0              ; initialize sse = 0
-
-loop
-    ; 1st 4 pixels
-    ldr     r5, [r0, #0x0]      ; load 4 src pixels
-    ldr     r6, [r2, #0x0]      ; load 4 ref pixels
-
-    mov     lr, #0              ; constant zero
-
-    usub8   r8, r5, r6          ; calculate difference
-    pld     [r0, r1, lsl #1]
-    sel     r7, r8, lr          ; select bytes with positive difference
-    usub8   r9, r6, r5          ; calculate difference with reversed operands
-    pld     [r2, r3, lsl #1]
-    sel     r8, r9, lr          ; select bytes with negative difference
-
-    ; calculate partial sums
-    usad8   r5, r7, lr          ; calculate sum of positive differences
-    usad8   r6, r8, lr          ; calculate sum of negative differences
-    orr     r8, r8, r7          ; differences of all 4 pixels
-
-    ldr     r5, [r0, #0x4]      ; load 4 src pixels
-
-    ; calculate sse
-    uxtb16  r6, r8              ; byte (two pixels) to halfwords
-    uxtb16  r7, r8, ror #8      ; another two pixels to halfwords
-    smlad   r4, r6, r6, r4      ; dual signed multiply, add and accumulate (1)
-
-    ; 2nd 4 pixels
-    ldr     r6, [r2, #0x4]      ; load 4 ref pixels
-    smlad   r4, r7, r7, r4      ; dual signed multiply, add and accumulate (2)
-
-    usub8   r8, r5, r6          ; calculate difference
-    sel     r7, r8, lr          ; select bytes with positive difference
-    usub8   r9, r6, r5          ; calculate difference with reversed operands
-    sel     r8, r9, lr          ; select bytes with negative difference
-
-    ; calculate partial sums
-    usad8   r5, r7, lr          ; calculate sum of positive differences
-    usad8   r6, r8, lr          ; calculate sum of negative differences
-    orr     r8, r8, r7          ; differences of all 4 pixels
-    ldr     r5, [r0, #0x8]      ; load 4 src pixels
-    ; calculate sse
-    uxtb16  r6, r8              ; byte (two pixels) to halfwords
-    uxtb16  r7, r8, ror #8      ; another two pixels to halfwords
-    smlad   r4, r6, r6, r4      ; dual signed multiply, add and accumulate (1)
-
-    ; 3rd 4 pixels
-    ldr     r6, [r2, #0x8]      ; load 4 ref pixels
-    smlad   r4, r7, r7, r4      ; dual signed multiply, add and accumulate (2)
-
-    usub8   r8, r5, r6          ; calculate difference
-    sel     r7, r8, lr          ; select bytes with positive difference
-    usub8   r9, r6, r5          ; calculate difference with reversed operands
-    sel     r8, r9, lr          ; select bytes with negative difference
-
-    ; calculate partial sums
-    usad8   r5, r7, lr          ; calculate sum of positive differences
-    usad8   r6, r8, lr          ; calculate sum of negative differences
-    orr     r8, r8, r7          ; differences of all 4 pixels
-
-    ldr     r5, [r0, #0xc]      ; load 4 src pixels
-
-    ; calculate sse
-    uxtb16  r6, r8              ; byte (two pixels) to halfwords
-    uxtb16  r7, r8, ror #8      ; another two pixels to halfwords
-    smlad   r4, r6, r6, r4      ; dual signed multiply, add and accumulate (1)
-
-    ; 4th 4 pixels
-    ldr     r6, [r2, #0xc]      ; load 4 ref pixels
-    smlad   r4, r7, r7, r4      ; dual signed multiply, add and accumulate (2)
-
-    usub8   r8, r5, r6          ; calculate difference
-    add     r0, r0, r1          ; set src_ptr to next row
-    sel     r7, r8, lr          ; select bytes with positive difference
-    usub8   r9, r6, r5          ; calculate difference with reversed operands
-    add     r2, r2, r3          ; set dst_ptr to next row
-    sel     r8, r9, lr          ; select bytes with negative difference
-
-    ; calculate partial sums
-    usad8   r5, r7, lr          ; calculate sum of positive differences
-    usad8   r6, r8, lr          ; calculate sum of negative differences
-    orr     r8, r8, r7          ; differences of all 4 pixels
-
-    subs    r12, r12, #1        ; next row
-
-    ; calculate sse
-    uxtb16  r6, r8              ; byte (two pixels) to halfwords
-    uxtb16  r7, r8, ror #8      ; another two pixels to halfwords
-    smlad   r4, r6, r6, r4      ; dual signed multiply, add and accumulate (1)
-    smlad   r4, r7, r7, r4      ; dual signed multiply, add and accumulate (2)
-
-    bne     loop
-
-    ; return stuff
-    ldr     r1, [sp, #28]       ; get address of sse
-    mov     r0, r4              ; return sse
-    str     r4, [r1]            ; store sse
-
-    pop     {r4-r9, pc}
-
-    ENDP
-
-    END
diff --git a/vp8/encoder/arm/neon/vp8_mse16x16_neon.c b/vp8/encoder/arm/neon/vp8_mse16x16_neon.c
deleted file mode 100644
index f806809df..000000000
--- a/vp8/encoder/arm/neon/vp8_mse16x16_neon.c
+++ /dev/null
@@ -1,131 +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>
-
-unsigned int vp8_mse16x16_neon(
-        const unsigned char *src_ptr,
-        int source_stride,
-        const unsigned char *ref_ptr,
-        int recon_stride,
-        unsigned int *sse) {
-    int i;
-    int16x4_t d22s16, d23s16, d24s16, d25s16, d26s16, d27s16, d28s16, d29s16;
-    int64x1_t d0s64;
-    uint8x16_t q0u8, q1u8, q2u8, q3u8;
-    int32x4_t q7s32, q8s32, q9s32, q10s32;
-    uint16x8_t q11u16, q12u16, q13u16, q14u16;
-    int64x2_t q1s64;
-
-    q7s32 = vdupq_n_s32(0);
-    q8s32 = vdupq_n_s32(0);
-    q9s32 = vdupq_n_s32(0);
-    q10s32 = vdupq_n_s32(0);
-
-    for (i = 0; i < 8; i++) {  // mse16x16_neon_loop
-        q0u8 = vld1q_u8(src_ptr);
-        src_ptr += source_stride;
-        q1u8 = vld1q_u8(src_ptr);
-        src_ptr += source_stride;
-        q2u8 = vld1q_u8(ref_ptr);
-        ref_ptr += recon_stride;
-        q3u8 = vld1q_u8(ref_ptr);
-        ref_ptr += recon_stride;
-
-        q11u16 = vsubl_u8(vget_low_u8(q0u8), vget_low_u8(q2u8));
-        q12u16 = vsubl_u8(vget_high_u8(q0u8), vget_high_u8(q2u8));
-        q13u16 = vsubl_u8(vget_low_u8(q1u8), vget_low_u8(q3u8));
-        q14u16 = vsubl_u8(vget_high_u8(q1u8), vget_high_u8(q3u8));
-
-        d22s16 = vreinterpret_s16_u16(vget_low_u16(q11u16));
-        d23s16 = vreinterpret_s16_u16(vget_high_u16(q11u16));
-        q7s32 = vmlal_s16(q7s32, d22s16, d22s16);
-        q8s32 = vmlal_s16(q8s32, d23s16, d23s16);
-
-        d24s16 = vreinterpret_s16_u16(vget_low_u16(q12u16));
-        d25s16 = vreinterpret_s16_u16(vget_high_u16(q12u16));
-        q9s32 = vmlal_s16(q9s32, d24s16, d24s16);
-        q10s32 = vmlal_s16(q10s32, d25s16, d25s16);
-
-        d26s16 = vreinterpret_s16_u16(vget_low_u16(q13u16));
-        d27s16 = vreinterpret_s16_u16(vget_high_u16(q13u16));
-        q7s32 = vmlal_s16(q7s32, d26s16, d26s16);
-        q8s32 = vmlal_s16(q8s32, d27s16, d27s16);
-
-        d28s16 = vreinterpret_s16_u16(vget_low_u16(q14u16));
-        d29s16 = vreinterpret_s16_u16(vget_high_u16(q14u16));
-        q9s32 = vmlal_s16(q9s32, d28s16, d28s16);
-        q10s32 = vmlal_s16(q10s32, d29s16, d29s16);
-    }
-
-    q7s32 = vaddq_s32(q7s32, q8s32);
-    q9s32 = vaddq_s32(q9s32, q10s32);
-    q10s32 = vaddq_s32(q7s32, q9s32);
-
-    q1s64 = vpaddlq_s32(q10s32);
-    d0s64 = vadd_s64(vget_low_s64(q1s64), vget_high_s64(q1s64));
-
-    vst1_lane_u32((uint32_t *)sse, vreinterpret_u32_s64(d0s64), 0);
-    return vget_lane_u32(vreinterpret_u32_s64(d0s64), 0);
-}
-
-unsigned int vp8_get4x4sse_cs_neon(
-        const unsigned char *src_ptr,
-        int source_stride,
-        const unsigned char *ref_ptr,
-        int recon_stride) {
-    int16x4_t d22s16, d24s16, d26s16, d28s16;
-    int64x1_t d0s64;
-    uint8x8_t d0u8, d1u8, d2u8, d3u8, d4u8, d5u8, d6u8, d7u8;
-    int32x4_t q7s32, q8s32, q9s32, q10s32;
-    uint16x8_t q11u16, q12u16, q13u16, q14u16;
-    int64x2_t q1s64;
-
-    d0u8 = vld1_u8(src_ptr);
-    src_ptr += source_stride;
-    d4u8 = vld1_u8(ref_ptr);
-    ref_ptr += recon_stride;
-    d1u8 = vld1_u8(src_ptr);
-    src_ptr += source_stride;
-    d5u8 = vld1_u8(ref_ptr);
-    ref_ptr += recon_stride;
-    d2u8 = vld1_u8(src_ptr);
-    src_ptr += source_stride;
-    d6u8 = vld1_u8(ref_ptr);
-    ref_ptr += recon_stride;
-    d3u8 = vld1_u8(src_ptr);
-    src_ptr += source_stride;
-    d7u8 = vld1_u8(ref_ptr);
-    ref_ptr += recon_stride;
-
-    q11u16 = vsubl_u8(d0u8, d4u8);
-    q12u16 = vsubl_u8(d1u8, d5u8);
-    q13u16 = vsubl_u8(d2u8, d6u8);
-    q14u16 = vsubl_u8(d3u8, d7u8);
-
-    d22s16 = vget_low_s16(vreinterpretq_s16_u16(q11u16));
-    d24s16 = vget_low_s16(vreinterpretq_s16_u16(q12u16));
-    d26s16 = vget_low_s16(vreinterpretq_s16_u16(q13u16));
-    d28s16 = vget_low_s16(vreinterpretq_s16_u16(q14u16));
-
-    q7s32 = vmull_s16(d22s16, d22s16);
-    q8s32 = vmull_s16(d24s16, d24s16);
-    q9s32 = vmull_s16(d26s16, d26s16);
-    q10s32 = vmull_s16(d28s16, d28s16);
-
-    q7s32 = vaddq_s32(q7s32, q8s32);
-    q9s32 = vaddq_s32(q9s32, q10s32);
-    q9s32 = vaddq_s32(q7s32, q9s32);
-
-    q1s64 = vpaddlq_s32(q9s32);
-    d0s64 = vadd_s64(vget_low_s64(q1s64), vget_high_s64(q1s64));
-
-    return vget_lane_u32(vreinterpret_u32_s64(d0s64), 0);
-}
diff --git a/vp8/encoder/encodeframe.c b/vp8/encoder/encodeframe.c
index 378e902c6..d381d8ddf 100644
--- a/vp8/encoder/encodeframe.c
+++ b/vp8/encoder/encodeframe.c
@@ -11,6 +11,7 @@
 
 #include "vpx_config.h"
 #include "vp8_rtcd.h"
+#include "./vpx_dsp_rtcd.h"
 #include "encodemb.h"
 #include "encodemv.h"
 #include "vp8/common/common.h"
@@ -90,7 +91,7 @@ static unsigned int tt_activity_measure( VP8_COMP *cpi, MACROBLOCK *x )
      *  lambda using a non-linear combination (e.g., the smallest, or second
      *  smallest, etc.).
      */
-    act =  vp8_variance16x16(x->src.y_buffer,
+    act =  vpx_variance16x16(x->src.y_buffer,
                     x->src.y_stride, VP8_VAR_OFFS, 0, &sse);
     act = act<<4;
 
diff --git a/vp8/encoder/encodeintra.c b/vp8/encoder/encodeintra.c
index cfa4cb927..e2de5eecb 100644
--- a/vp8/encoder/encodeintra.c
+++ b/vp8/encoder/encodeintra.c
@@ -11,6 +11,7 @@
 
 #include "vpx_config.h"
 #include "vp8_rtcd.h"
+#include "./vpx_dsp_rtcd.h"
 #include "quantize.h"
 #include "vp8/common/reconintra4x4.h"
 #include "encodemb.h"
@@ -44,7 +45,7 @@ int vp8_encode_intra(VP8_COMP *cpi, MACROBLOCK *x, int use_dc_pred)
         }
     }
 
-    intra_pred_var = vp8_get_mb_ss(x->src_diff);
+    intra_pred_var = vpx_get_mb_ss(x->src_diff);
 
     return intra_pred_var;
 }
diff --git a/vp8/encoder/firstpass.c b/vp8/encoder/firstpass.c
index a6ff0e7a0..3deb4abb3 100644
--- a/vp8/encoder/firstpass.c
+++ b/vp8/encoder/firstpass.c
@@ -12,6 +12,7 @@
 #include <limits.h>
 #include <stdio.h>
 
+#include "./vpx_dsp_rtcd.h"
 #include "./vpx_scale_rtcd.h"
 #include "block.h"
 #include "onyx_int.h"
@@ -422,14 +423,14 @@ static void zz_motion_search( VP8_COMP *cpi, MACROBLOCK * x,
     /* Set up pointers for this macro block raw buffer */
     raw_ptr = (unsigned char *)(raw_buffer->y_buffer + recon_yoffset
                                 + d->offset);
-    vp8_mse16x16 ( src_ptr, src_stride, raw_ptr, raw_stride,
-                   (unsigned int *)(raw_motion_err));
+    vpx_mse16x16(src_ptr, src_stride, raw_ptr, raw_stride,
+                 (unsigned int *)(raw_motion_err));
 
     /* Set up pointers for this macro block recon buffer */
     xd->pre.y_buffer = recon_buffer->y_buffer + recon_yoffset;
     ref_ptr = (unsigned char *)(xd->pre.y_buffer + d->offset );
-    vp8_mse16x16 ( src_ptr, src_stride, ref_ptr, ref_stride,
-                   (unsigned int *)(best_motion_err));
+    vpx_mse16x16(src_ptr, src_stride, ref_ptr, ref_stride,
+                 (unsigned int *)(best_motion_err));
 }
 
 static void first_pass_motion_search(VP8_COMP *cpi, MACROBLOCK *x,
@@ -453,7 +454,7 @@ static void first_pass_motion_search(VP8_COMP *cpi, MACROBLOCK *x,
     int new_mv_mode_penalty = 256;
 
     /* override the default variance function to use MSE */
-    v_fn_ptr.vf    = vp8_mse16x16;
+    v_fn_ptr.vf    = vpx_mse16x16;
 
     /* Set up pointers for this macro block recon buffer */
     xd->pre.y_buffer = recon_buffer->y_buffer + recon_yoffset;
diff --git a/vp8/encoder/onyx_if.c b/vp8/encoder/onyx_if.c
index c2bb23295..40e29e191 100644
--- a/vp8/encoder/onyx_if.c
+++ b/vp8/encoder/onyx_if.c
@@ -2131,7 +2131,7 @@ struct VP8_COMP* vp8_create_compressor(VP8_CONFIG *oxcf)
 #endif
 
     cpi->fn_ptr[BLOCK_16X16].sdf            = vpx_sad16x16;
-    cpi->fn_ptr[BLOCK_16X16].vf             = vp8_variance16x16;
+    cpi->fn_ptr[BLOCK_16X16].vf             = vpx_variance16x16;
     cpi->fn_ptr[BLOCK_16X16].svf            = vp8_sub_pixel_variance16x16;
     cpi->fn_ptr[BLOCK_16X16].svf_halfpix_h  = vp8_variance_halfpixvar16x16_h;
     cpi->fn_ptr[BLOCK_16X16].svf_halfpix_v  = vp8_variance_halfpixvar16x16_v;
@@ -2141,7 +2141,7 @@ struct VP8_COMP* vp8_create_compressor(VP8_CONFIG *oxcf)
     cpi->fn_ptr[BLOCK_16X16].sdx4df         = vpx_sad16x16x4d;
 
     cpi->fn_ptr[BLOCK_16X8].sdf            = vpx_sad16x8;
-    cpi->fn_ptr[BLOCK_16X8].vf             = vp8_variance16x8;
+    cpi->fn_ptr[BLOCK_16X8].vf             = vpx_variance16x8;
     cpi->fn_ptr[BLOCK_16X8].svf            = vp8_sub_pixel_variance16x8;
     cpi->fn_ptr[BLOCK_16X8].svf_halfpix_h  = NULL;
     cpi->fn_ptr[BLOCK_16X8].svf_halfpix_v  = NULL;
@@ -2151,7 +2151,7 @@ struct VP8_COMP* vp8_create_compressor(VP8_CONFIG *oxcf)
     cpi->fn_ptr[BLOCK_16X8].sdx4df         = vpx_sad16x8x4d;
 
     cpi->fn_ptr[BLOCK_8X16].sdf            = vpx_sad8x16;
-    cpi->fn_ptr[BLOCK_8X16].vf             = vp8_variance8x16;
+    cpi->fn_ptr[BLOCK_8X16].vf             = vpx_variance8x16;
     cpi->fn_ptr[BLOCK_8X16].svf            = vp8_sub_pixel_variance8x16;
     cpi->fn_ptr[BLOCK_8X16].svf_halfpix_h  = NULL;
     cpi->fn_ptr[BLOCK_8X16].svf_halfpix_v  = NULL;
@@ -2161,7 +2161,7 @@ struct VP8_COMP* vp8_create_compressor(VP8_CONFIG *oxcf)
     cpi->fn_ptr[BLOCK_8X16].sdx4df         = vpx_sad8x16x4d;
 
     cpi->fn_ptr[BLOCK_8X8].sdf            = vpx_sad8x8;
-    cpi->fn_ptr[BLOCK_8X8].vf             = vp8_variance8x8;
+    cpi->fn_ptr[BLOCK_8X8].vf             = vpx_variance8x8;
     cpi->fn_ptr[BLOCK_8X8].svf            = vp8_sub_pixel_variance8x8;
     cpi->fn_ptr[BLOCK_8X8].svf_halfpix_h  = NULL;
     cpi->fn_ptr[BLOCK_8X8].svf_halfpix_v  = NULL;
@@ -2171,7 +2171,7 @@ struct VP8_COMP* vp8_create_compressor(VP8_CONFIG *oxcf)
     cpi->fn_ptr[BLOCK_8X8].sdx4df         = vpx_sad8x8x4d;
 
     cpi->fn_ptr[BLOCK_4X4].sdf            = vpx_sad4x4;
-    cpi->fn_ptr[BLOCK_4X4].vf             = vp8_variance4x4;
+    cpi->fn_ptr[BLOCK_4X4].vf             = vpx_variance4x4;
     cpi->fn_ptr[BLOCK_4X4].svf            = vp8_sub_pixel_variance4x4;
     cpi->fn_ptr[BLOCK_4X4].svf_halfpix_h  = NULL;
     cpi->fn_ptr[BLOCK_4X4].svf_halfpix_v  = NULL;
@@ -2558,7 +2558,7 @@ static uint64_t calc_plane_error(unsigned char *orig, int orig_stride,
         {
             unsigned int sse;
 
-            vp8_mse16x16(orig + col, orig_stride,
+            vpx_mse16x16(orig + col, orig_stride,
                                             recon + col, recon_stride,
                                             &sse);
             total_sse += sse;
@@ -3384,7 +3384,7 @@ static int measure_square_diff_partial(YV12_BUFFER_CONFIG *source,
                 int index = block_index_row + (j >> 4);
                 if (cpi->consec_zero_last[index] >= min_consec_zero_last) {
                   unsigned int sse;
-                  Total += vp8_mse16x16(src + j,
+                  Total += vpx_mse16x16(src + j,
                                         source->y_stride,
                                         dst + j, dest->y_stride,
                                         &sse);
@@ -3448,7 +3448,7 @@ static void process_denoiser_mode_change(VP8_COMP *cpi) {
       int index = block_index_row + (j >> 4);
       if (cpi->consec_zero_last[index] >= min_consec_zero_last) {
         unsigned int sse;
-        const unsigned int var = vp8_variance16x16(src + j,
+        const unsigned int var = vpx_variance16x16(src + j,
                                                    ystride,
                                                    dst + j,
                                                    ystride,
@@ -3458,7 +3458,7 @@ static void process_denoiser_mode_change(VP8_COMP *cpi) {
         // is small (to avoid effects from lighting change).
         if ((sse - var) < 128) {
           unsigned int sse2;
-          const unsigned int act = vp8_variance16x16(src + j,
+          const unsigned int act = vpx_variance16x16(src + j,
                                                      ystride,
                                                      const_source,
                                                      0,
@@ -5993,7 +5993,8 @@ int vp8_calc_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest)
         for (j = 0; j < source->y_width; j += 16)
         {
             unsigned int sse;
-            Total += vp8_mse16x16(src + j, source->y_stride, dst + j, dest->y_stride, &sse);
+            Total += vpx_mse16x16(src + j, source->y_stride,
+                                  dst + j, dest->y_stride, &sse);
         }
 
         src += 16 * source->y_stride;
diff --git a/vp8/encoder/pickinter.c b/vp8/encoder/pickinter.c
index 98ea5a040..053bf119a 100644
--- a/vp8/encoder/pickinter.c
+++ b/vp8/encoder/pickinter.c
@@ -11,6 +11,7 @@
 
 #include <limits.h>
 #include "vpx_config.h"
+#include "./vpx_dsp_rtcd.h"
 #include "onyx_int.h"
 #include "modecosts.h"
 #include "encodeintra.h"
@@ -215,33 +216,6 @@ int vp8_get_inter_mbpred_error(MACROBLOCK *mb,
 
 }
 
-
-unsigned int vp8_get4x4sse_cs_c
-(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride
-)
-{
-    int distortion = 0;
-    int r, c;
-
-    for (r = 0; r < 4; r++)
-    {
-        for (c = 0; c < 4; c++)
-        {
-            int diff = src_ptr[c] - ref_ptr[c];
-            distortion += diff * diff;
-        }
-
-        src_ptr += source_stride;
-        ref_ptr += recon_stride;
-    }
-
-    return distortion;
-}
-
 static int get_prediction_error(BLOCK *be, BLOCKD *b)
 {
     unsigned char *sptr;
@@ -249,7 +223,7 @@ static int get_prediction_error(BLOCK *be, BLOCKD *b)
     sptr = (*(be->base_src) + be->src);
     dptr = b->predictor;
 
-    return vp8_get4x4sse_cs(sptr, be->src_stride, dptr, 16);
+    return vpx_get4x4sse_cs(sptr, be->src_stride, dptr, 16);
 
 }
 
@@ -1037,7 +1011,7 @@ void vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset,
             else
             {
                 rate2 += rate;
-                distortion2 = vp8_variance16x16(
+                distortion2 = vpx_variance16x16(
                                     *(b->base_src), b->src_stride,
                                     x->e_mbd.predictor, 16, &sse);
                 this_rd = RDCOST(x->rdmult, x->rddiv, rate2, distortion2);
@@ -1066,7 +1040,7 @@ void vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset,
                                              xd->dst.y_stride,
                                              xd->predictor,
                                              16);
-            distortion2 = vp8_variance16x16
+            distortion2 = vpx_variance16x16
                                           (*(b->base_src), b->src_stride,
                                           x->e_mbd.predictor, 16, &sse);
             rate2 += x->mbmode_cost[x->e_mbd.frame_type][x->e_mbd.mode_info_context->mbmi.mode];
@@ -1547,7 +1521,7 @@ void vp8_pick_intra_mode(MACROBLOCK *x, int *rate_)
                                          xd->dst.y_stride,
                                          xd->predictor,
                                          16);
-        distortion = vp8_variance16x16
+        distortion = vpx_variance16x16
             (*(b->base_src), b->src_stride, xd->predictor, 16, &sse);
         rate = x->mbmode_cost[xd->frame_type][mode];
         this_rd = RDCOST(x->rdmult, x->rddiv, rate, distortion);
diff --git a/vp8/encoder/picklpf.c b/vp8/encoder/picklpf.c
index 890053dcf..875b37f68 100644
--- a/vp8/encoder/picklpf.c
+++ b/vp8/encoder/picklpf.c
@@ -9,6 +9,7 @@
  */
 
 
+#include "./vpx_dsp_rtcd.h"
 #include "./vpx_scale_rtcd.h"
 #include "vp8/common/onyxc_int.h"
 #include "onyx_int.h"
@@ -83,7 +84,7 @@ static int calc_partial_ssl_err(YV12_BUFFER_CONFIG *source,
         for (j = 0; j < source->y_width; j += 16)
         {
             unsigned int sse;
-            Total += vp8_mse16x16(src + j, source->y_stride,
+            Total += vpx_mse16x16(src + j, source->y_stride,
                                                      dst + j, dest->y_stride,
                                                      &sse);
         }
diff --git a/vp8/encoder/rdopt.c b/vp8/encoder/rdopt.c
index 9ccd85eb9..17194f0d4 100644
--- a/vp8/encoder/rdopt.c
+++ b/vp8/encoder/rdopt.c
@@ -15,6 +15,7 @@
 #include <assert.h>
 #include "vpx_config.h"
 #include "vp8_rtcd.h"
+#include "./vpx_dsp_rtcd.h"
 #include "tokenize.h"
 #include "treewriter.h"
 #include "onyx_int.h"
@@ -507,9 +508,9 @@ int VP8_UVSSE(MACROBLOCK *x)
     }
     else
     {
-        vp8_variance8x8(uptr, pre_stride,
+        vpx_variance8x8(uptr, pre_stride,
             upred_ptr, uv_stride, &sse2);
-        vp8_variance8x8(vptr, pre_stride,
+        vpx_variance8x8(vptr, pre_stride,
             vpred_ptr, uv_stride, &sse1);
         sse2 += sse1;
     }
@@ -1783,7 +1784,7 @@ static int evaluate_inter_mode_rd(int mdcounts[4],
         if(threshold < x->encode_breakout)
             threshold = x->encode_breakout;
 
-        var = vp8_variance16x16
+        var = vpx_variance16x16
                 (*(b->base_src), b->src_stride,
                 x->e_mbd.predictor, 16, &sse);
 
diff --git a/vp8/vp8_common.mk b/vp8/vp8_common.mk
index b4c814075..c71d592f5 100644
--- a/vp8/vp8_common.mk
+++ b/vp8/vp8_common.mk
@@ -145,8 +145,6 @@ VP8_COMMON_SRCS-$(HAVE_MEDIA)  += common/arm/armv6/intra4x4_predict_v6$(ASM)
 VP8_COMMON_SRCS-$(HAVE_MEDIA)  += common/arm/armv6/dequant_idct_v6$(ASM)
 VP8_COMMON_SRCS-$(HAVE_MEDIA)  += common/arm/armv6/dequantize_v6$(ASM)
 VP8_COMMON_SRCS-$(HAVE_MEDIA)  += common/arm/armv6/idct_blk_v6.c
-VP8_COMMON_SRCS-$(HAVE_MEDIA)  += common/arm/armv6/vp8_variance8x8_armv6$(ASM)
-VP8_COMMON_SRCS-$(HAVE_MEDIA)  += common/arm/armv6/vp8_variance16x16_armv6$(ASM)
 VP8_COMMON_SRCS-$(HAVE_MEDIA)  += common/arm/armv6/vp8_variance_halfpixvar16x16_h_armv6$(ASM)
 VP8_COMMON_SRCS-$(HAVE_MEDIA)  += common/arm/armv6/vp8_variance_halfpixvar16x16_v_armv6$(ASM)
 VP8_COMMON_SRCS-$(HAVE_MEDIA)  += common/arm/armv6/vp8_variance_halfpixvar16x16_hv_armv6$(ASM)
@@ -168,7 +166,6 @@ VP8_COMMON_SRCS-$(HAVE_NEON)  += common/arm/neon/mbloopfilter_neon.c
 VP8_COMMON_SRCS-$(HAVE_NEON)  += common/arm/neon/reconintra_neon.c
 VP8_COMMON_SRCS-$(HAVE_NEON)  += common/arm/neon/shortidct4x4llm_neon.c
 VP8_COMMON_SRCS-$(HAVE_NEON)  += common/arm/neon/sixtappredict_neon.c
-VP8_COMMON_SRCS-$(HAVE_NEON)  += common/arm/neon/variance_neon.c
 VP8_COMMON_SRCS-$(HAVE_NEON)  += common/arm/neon/vp8_subpixelvariance_neon.c
 
 $(eval $(call rtcd_h_template,vp8_rtcd,vp8/common/rtcd_defs.pl))
diff --git a/vp8/vp8cx_arm.mk b/vp8/vp8cx_arm.mk
index 050030179..0b0f6a70a 100644
--- a/vp8/vp8cx_arm.mk
+++ b/vp8/vp8cx_arm.mk
@@ -18,7 +18,6 @@ VP8_CX_SRCS-$(ARCH_ARM)  += encoder/arm/dct_arm.c
 #File list for media
 # encoder
 VP8_CX_SRCS-$(HAVE_MEDIA)  += encoder/arm/armv6/vp8_short_fdct4x4_armv6$(ASM)
-VP8_CX_SRCS-$(HAVE_MEDIA)  += encoder/arm/armv6/vp8_mse16x16_armv6$(ASM)
 VP8_CX_SRCS-$(HAVE_MEDIA)  += encoder/arm/armv6/walsh_v6$(ASM)
 
 #File list for neon
@@ -27,5 +26,4 @@ VP8_CX_SRCS-$(HAVE_NEON)  += encoder/arm/neon/denoising_neon.c
 VP8_CX_SRCS-$(HAVE_NEON)  += encoder/arm/neon/fastquantizeb_neon.c
 VP8_CX_SRCS-$(HAVE_NEON)  += encoder/arm/neon/shortfdct_neon.c
 VP8_CX_SRCS-$(HAVE_NEON)  += encoder/arm/neon/subtract_neon.c
-VP8_CX_SRCS-$(HAVE_NEON)  += encoder/arm/neon/vp8_mse16x16_neon.c
 VP8_CX_SRCS-$(HAVE_NEON)  += encoder/arm/neon/vp8_shortwalsh4x4_neon.c