Move intra prediction functions from vp9/common/ to vpx_dsp/

Change-Id: I64edc26cf4aab050c83f2d393df6250628ad43b8

3 files changed, 0 insertions, 2179 deletions

diff --git a/vp9/common/x86/vp9_high_intrapred_sse2.asm b/vp9/common/x86/vp9_high_intrapred_sse2.asm
deleted file mode 100644
index b12d29c0a..000000000
--- a/vp9/common/x86/vp9_high_intrapred_sse2.asm
+++ /dev/null
@@ -1,476 +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 "third_party/x86inc/x86inc.asm"
-
-SECTION_RODATA
-pw_4:  times 8 dw 4
-pw_8:  times 8 dw 8
-pw_16: times 4 dd 16
-pw_32: times 4 dd 32
-
-SECTION .text
-INIT_MMX sse
-cglobal highbd_dc_predictor_4x4, 4, 5, 4, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  movq                  m0, [aboveq]
-  movq                  m2, [leftq]
-  DEFINE_ARGS dst, stride, one
-  mov                 oned, 0x0001
-  pxor                  m1, m1
-  movd                  m3, oned
-  pshufw                m3, m3, 0x0
-  paddw                 m0, m2
-  pmaddwd               m0, m3
-  packssdw              m0, m1
-  pmaddwd               m0, m3
-  paddw                 m0, [GLOBAL(pw_4)]
-  psraw                 m0, 3
-  pshufw                m0, m0, 0x0
-  movq    [dstq          ], m0
-  movq    [dstq+strideq*2], m0
-  lea                 dstq, [dstq+strideq*4]
-  movq    [dstq          ], m0
-  movq    [dstq+strideq*2], m0
-
-  RESTORE_GOT
-  RET
-
-INIT_XMM sse2
-cglobal highbd_dc_predictor_8x8, 4, 5, 4, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  mova                  m0, [aboveq]
-  mova                  m2, [leftq]
-  DEFINE_ARGS dst, stride, stride3, one
-  mov                 oned, 0x00010001
-  lea             stride3q, [strideq*3]
-  movd                  m3, oned
-  pshufd                m3, m3, 0x0
-  paddw                 m0, m2
-  pmaddwd               m0, m3
-  packssdw              m0, m1
-  pmaddwd               m0, m3
-  packssdw              m0, m1
-  pmaddwd               m0, m3
-  paddw                 m0, [GLOBAL(pw_8)]
-  psrlw                 m0, 4
-  pshuflw               m0, m0, 0x0
-  punpcklqdq            m0, m0
-  mova   [dstq           ], m0
-  mova   [dstq+strideq*2 ], m0
-  mova   [dstq+strideq*4 ], m0
-  mova   [dstq+stride3q*2], m0
-  lea                 dstq, [dstq+strideq*8]
-  mova   [dstq           ], m0
-  mova   [dstq+strideq*2 ], m0
-  mova   [dstq+strideq*4 ], m0
-  mova   [dstq+stride3q*2], m0
-
-  RESTORE_GOT
-  RET
-
-INIT_XMM sse2
-cglobal highbd_dc_predictor_16x16, 4, 5, 5, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  mova                  m0, [aboveq]
-  mova                  m3, [aboveq+16]
-  mova                  m2, [leftq]
-  mova                  m4, [leftq+16]
-  DEFINE_ARGS dst, stride, stride3, lines4
-  lea             stride3q, [strideq*3]
-  mov              lines4d, 4
-  paddw                 m0, m2
-  paddw                 m0, m3
-  paddw                 m0, m4
-  movhlps               m2, m0
-  paddw                 m0, m2
-  punpcklwd             m0, m1
-  movhlps               m2, m0
-  paddd                 m0, m2
-  punpckldq             m0, m1
-  movhlps               m2, m0
-  paddd                 m0, m2
-  paddd                 m0, [GLOBAL(pw_16)]
-  psrad                 m0, 5
-  pshuflw               m0, m0, 0x0
-  punpcklqdq            m0, m0
-.loop:
-  mova   [dstq              ], m0
-  mova   [dstq           +16], m0
-  mova   [dstq+strideq*2    ], m0
-  mova   [dstq+strideq*2 +16], m0
-  mova   [dstq+strideq*4    ], m0
-  mova   [dstq+strideq*4 +16], m0
-  mova   [dstq+stride3q*2   ], m0
-  mova   [dstq+stride3q*2+16], m0
-  lea                 dstq, [dstq+strideq*8]
-  dec              lines4d
-  jnz .loop
-
-  RESTORE_GOT
-  REP_RET
-
-%if ARCH_X86_64
-INIT_XMM sse2
-cglobal highbd_dc_predictor_32x32, 4, 5, 9, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  mova                  m0, [aboveq]
-  mova                  m2, [aboveq+16]
-  mova                  m3, [aboveq+32]
-  mova                  m4, [aboveq+48]
-  mova                  m5, [leftq]
-  mova                  m6, [leftq+16]
-  mova                  m7, [leftq+32]
-  mova                  m8, [leftq+48]
-  DEFINE_ARGS dst, stride, stride3, lines4
-  lea             stride3q, [strideq*3]
-  mov              lines4d, 8
-  paddw                 m0, m2
-  paddw                 m0, m3
-  paddw                 m0, m4
-  paddw                 m0, m5
-  paddw                 m0, m6
-  paddw                 m0, m7
-  paddw                 m0, m8
-  movhlps               m2, m0
-  paddw                 m0, m2
-  punpcklwd             m0, m1
-  movhlps               m2, m0
-  paddd                 m0, m2
-  punpckldq             m0, m1
-  movhlps               m2, m0
-  paddd                 m0, m2
-  paddd                 m0, [GLOBAL(pw_32)]
-  psrad                 m0, 6
-  pshuflw               m0, m0, 0x0
-  punpcklqdq            m0, m0
-.loop:
-  mova [dstq               ], m0
-  mova [dstq          +16  ], m0
-  mova [dstq          +32  ], m0
-  mova [dstq          +48  ], m0
-  mova [dstq+strideq*2     ], m0
-  mova [dstq+strideq*2+16  ], m0
-  mova [dstq+strideq*2+32  ], m0
-  mova [dstq+strideq*2+48  ], m0
-  mova [dstq+strideq*4     ], m0
-  mova [dstq+strideq*4+16  ], m0
-  mova [dstq+strideq*4+32  ], m0
-  mova [dstq+strideq*4+48  ], m0
-  mova [dstq+stride3q*2    ], m0
-  mova [dstq+stride3q*2 +16], m0
-  mova [dstq+stride3q*2 +32], m0
-  mova [dstq+stride3q*2 +48], m0
-  lea                 dstq, [dstq+strideq*8]
-  dec              lines4d
-  jnz .loop
-
-  RESTORE_GOT
-  REP_RET
-%endif
-
-INIT_MMX sse
-cglobal highbd_v_predictor_4x4, 3, 3, 1, dst, stride, above
-  movq                  m0, [aboveq]
-  movq    [dstq          ], m0
-  movq    [dstq+strideq*2], m0
-  lea                 dstq, [dstq+strideq*4]
-  movq    [dstq          ], m0
-  movq    [dstq+strideq*2], m0
-  RET
-
-INIT_XMM sse2
-cglobal highbd_v_predictor_8x8, 3, 3, 1, dst, stride, above
-  mova                  m0, [aboveq]
-  DEFINE_ARGS dst, stride, stride3
-  lea             stride3q, [strideq*3]
-  mova   [dstq           ], m0
-  mova   [dstq+strideq*2 ], m0
-  mova   [dstq+strideq*4 ], m0
-  mova   [dstq+stride3q*2], m0
-  lea                 dstq, [dstq+strideq*8]
-  mova   [dstq           ], m0
-  mova   [dstq+strideq*2 ], m0
-  mova   [dstq+strideq*4 ], m0
-  mova   [dstq+stride3q*2], m0
-  RET
-
-INIT_XMM sse2
-cglobal highbd_v_predictor_16x16, 3, 4, 2, dst, stride, above
-  mova                  m0, [aboveq]
-  mova                  m1, [aboveq+16]
-  DEFINE_ARGS dst, stride, stride3, nlines4
-  lea             stride3q, [strideq*3]
-  mov              nlines4d, 4
-.loop:
-  mova    [dstq              ], m0
-  mova    [dstq           +16], m1
-  mova    [dstq+strideq*2    ], m0
-  mova    [dstq+strideq*2 +16], m1
-  mova    [dstq+strideq*4    ], m0
-  mova    [dstq+strideq*4 +16], m1
-  mova    [dstq+stride3q*2   ], m0
-  mova    [dstq+stride3q*2+16], m1
-  lea                 dstq, [dstq+strideq*8]
-  dec             nlines4d
-  jnz .loop
-  REP_RET
-
-INIT_XMM sse2
-cglobal highbd_v_predictor_32x32, 3, 4, 4, dst, stride, above
-  mova                  m0, [aboveq]
-  mova                  m1, [aboveq+16]
-  mova                  m2, [aboveq+32]
-  mova                  m3, [aboveq+48]
-  DEFINE_ARGS dst, stride, stride3, nlines4
-  lea             stride3q, [strideq*3]
-  mov              nlines4d, 8
-.loop:
-  mova [dstq               ], m0
-  mova [dstq            +16], m1
-  mova [dstq            +32], m2
-  mova [dstq            +48], m3
-  mova [dstq+strideq*2     ], m0
-  mova [dstq+strideq*2  +16], m1
-  mova [dstq+strideq*2  +32], m2
-  mova [dstq+strideq*2  +48], m3
-  mova [dstq+strideq*4     ], m0
-  mova [dstq+strideq*4  +16], m1
-  mova [dstq+strideq*4  +32], m2
-  mova [dstq+strideq*4  +48], m3
-  mova [dstq+stride3q*2    ], m0
-  mova [dstq+stride3q*2 +16], m1
-  mova [dstq+stride3q*2 +32], m2
-  mova [dstq+stride3q*2 +48], m3
-  lea                 dstq, [dstq+strideq*8]
-  dec             nlines4d
-  jnz .loop
-  REP_RET
-
-INIT_MMX sse
-cglobal highbd_tm_predictor_4x4, 5, 6, 5, dst, stride, above, left, bps, one
-  movd                  m1, [aboveq-2]
-  movq                  m0, [aboveq]
-  pshufw                m1, m1, 0x0
-  ; Get the values to compute the maximum value at this bit depth
-  mov                 oned, 1
-  movd                  m3, oned
-  movd                  m4, bpsd
-  pshufw                m3, m3, 0x0
-  DEFINE_ARGS dst, stride, line, left
-  mov                lineq, -2
-  mova                  m2, m3
-  psllw                 m3, m4
-  add                leftq, 8
-  psubw                 m3, m2 ; max possible value
-  pxor                  m4, m4 ; min possible value
-  psubw                 m0, m1
-.loop:
-  movq                  m1, [leftq+lineq*4]
-  movq                  m2, [leftq+lineq*4+2]
-  pshufw                m1, m1, 0x0
-  pshufw                m2, m2, 0x0
-  paddw                 m1, m0
-  paddw                 m2, m0
-  ;Clamp to the bit-depth
-  pminsw                m1, m3
-  pminsw                m2, m3
-  pmaxsw                m1, m4
-  pmaxsw                m2, m4
-  ;Store the values
-  movq    [dstq          ], m1
-  movq    [dstq+strideq*2], m2
-  lea                 dstq, [dstq+strideq*4]
-  inc                lineq
-  jnz .loop
-  REP_RET
-
-INIT_XMM sse2
-cglobal highbd_tm_predictor_8x8, 5, 6, 5, dst, stride, above, left, bps, one
-  movd                  m1, [aboveq-2]
-  mova                  m0, [aboveq]
-  pshuflw               m1, m1, 0x0
-  ; Get the values to compute the maximum value at this bit depth
-  mov                 oned, 1
-  pxor                  m3, m3
-  pxor                  m4, m4
-  pinsrw                m3, oned, 0
-  pinsrw                m4, bpsd, 0
-  pshuflw               m3, m3, 0x0
-  DEFINE_ARGS dst, stride, line, left
-  punpcklqdq            m3, m3
-  mov                lineq, -4
-  mova                  m2, m3
-  punpcklqdq            m1, m1
-  psllw                 m3, m4
-  add                leftq, 16
-  psubw                 m3, m2 ; max possible value
-  pxor                  m4, m4 ; min possible value
-  psubw                 m0, m1
-.loop:
-  movd                  m1, [leftq+lineq*4]
-  movd                  m2, [leftq+lineq*4+2]
-  pshuflw               m1, m1, 0x0
-  pshuflw               m2, m2, 0x0
-  punpcklqdq            m1, m1
-  punpcklqdq            m2, m2
-  paddw                 m1, m0
-  paddw                 m2, m0
-  ;Clamp to the bit-depth
-  pminsw                m1, m3
-  pminsw                m2, m3
-  pmaxsw                m1, m4
-  pmaxsw                m2, m4
-  ;Store the values
-  mova      [dstq          ], m1
-  mova      [dstq+strideq*2], m2
-  lea                 dstq, [dstq+strideq*4]
-  inc                lineq
-  jnz .loop
-  REP_RET
-
-%if ARCH_X86_64
-INIT_XMM sse2
-cglobal highbd_tm_predictor_16x16, 5, 6, 9, dst, stride, above, left, bps, one
-  movd                  m2, [aboveq-2]
-  mova                  m0, [aboveq]
-  mova                  m1, [aboveq+16]
-  pshuflw               m2, m2, 0x0
-  ; Get the values to compute the maximum value at this bit depth
-  mov                 oned, 1
-  pxor                  m7, m7
-  pxor                  m8, m8
-  pinsrw                m7, oned, 0
-  pinsrw                m8, bpsd, 0
-  pshuflw               m7, m7, 0x0
-  DEFINE_ARGS dst, stride, line, left
-  punpcklqdq            m7, m7
-  mov                lineq, -8
-  mova                  m5, m7
-  punpcklqdq            m2, m2
-  psllw                 m7, m8
-  add                leftq, 32
-  psubw                 m7, m5 ; max possible value
-  pxor                  m8, m8 ; min possible value
-  psubw                 m0, m2
-  psubw                 m1, m2
-.loop:
-  movd                  m2, [leftq+lineq*4]
-  movd                  m3, [leftq+lineq*4+2]
-  pshuflw               m2, m2, 0x0
-  pshuflw               m3, m3, 0x0
-  punpcklqdq            m2, m2
-  punpcklqdq            m3, m3
-  paddw                 m4, m2, m0
-  paddw                 m5, m3, m0
-  paddw                 m2, m1
-  paddw                 m3, m1
-  ;Clamp to the bit-depth
-  pminsw                m4, m7
-  pminsw                m5, m7
-  pminsw                m2, m7
-  pminsw                m3, m7
-  pmaxsw                m4, m8
-  pmaxsw                m5, m8
-  pmaxsw                m2, m8
-  pmaxsw                m3, m8
-  ;Store the values
-  mova   [dstq             ], m4
-  mova   [dstq+strideq*2   ], m5
-  mova   [dstq          +16], m2
-  mova   [dstq+strideq*2+16], m3
-  lea                 dstq, [dstq+strideq*4]
-  inc                lineq
-  jnz .loop
-  REP_RET
-
-INIT_XMM sse2
-cglobal highbd_tm_predictor_32x32, 5, 6, 12, dst, stride, above, left, bps, one
-  movd                  m0, [aboveq-2]
-  mova                  m1, [aboveq]
-  mova                  m2, [aboveq+16]
-  mova                  m3, [aboveq+32]
-  mova                  m4, [aboveq+48]
-  pshuflw               m0, m0, 0x0
-  ; Get the values to compute the maximum value at this bit depth
-  mov                 oned, 1
-  pxor                 m10, m10
-  pxor                 m11, m11
-  pinsrw               m10, oned, 0
-  pinsrw               m11, bpsd, 0
-  pshuflw              m10, m10, 0x0
-  DEFINE_ARGS dst, stride, line, left
-  punpcklqdq           m10, m10
-  mov                lineq, -16
-  mova                  m5, m10
-  punpcklqdq            m0, m0
-  psllw                m10, m11
-  add                leftq, 64
-  psubw                m10, m5 ; max possible value
-  pxor                 m11, m11 ; min possible value
-  psubw                 m1, m0
-  psubw                 m2, m0
-  psubw                 m3, m0
-  psubw                 m4, m0
-.loop:
-  movd                  m5, [leftq+lineq*4]
-  movd                  m6, [leftq+lineq*4+2]
-  pshuflw               m5, m5, 0x0
-  pshuflw               m6, m6, 0x0
-  punpcklqdq            m5, m5
-  punpcklqdq            m6, m6
-  paddw                 m7, m5, m1
-  paddw                 m8, m5, m2
-  paddw                 m9, m5, m3
-  paddw                 m5, m4
-  ;Clamp these values to the bit-depth
-  pminsw                m7, m10
-  pminsw                m8, m10
-  pminsw                m9, m10
-  pminsw                m5, m10
-  pmaxsw                m7, m11
-  pmaxsw                m8, m11
-  pmaxsw                m9, m11
-  pmaxsw                m5, m11
-  ;Store these values
-  mova   [dstq           ], m7
-  mova   [dstq        +16], m8
-  mova   [dstq        +32], m9
-  mova   [dstq        +48], m5
-  paddw                 m7, m6, m1
-  paddw                 m8, m6, m2
-  paddw                 m9, m6, m3
-  paddw                 m6, m4
-  ;Clamp these values to the bit-depth
-  pminsw                m7, m10
-  pminsw                m8, m10
-  pminsw                m9, m10
-  pminsw                m6, m10
-  pmaxsw                m7, m11
-  pmaxsw                m8, m11
-  pmaxsw                m9, m11
-  pmaxsw                m6, m11
-  ;Store these values
-  mova   [dstq+strideq*2   ], m7
-  mova   [dstq+strideq*2+16], m8
-  mova   [dstq+strideq*2+32], m9
-  mova   [dstq+strideq*2+48], m6
-  lea                 dstq, [dstq+strideq*4]
-  inc                lineq
-  jnz .loop
-  REP_RET
-%endif
diff --git a/vp9/common/x86/vp9_intrapred_sse2.asm b/vp9/common/x86/vp9_intrapred_sse2.asm
deleted file mode 100644
index 22b573188..000000000
--- a/vp9/common/x86/vp9_intrapred_sse2.asm
+++ /dev/null
@@ -1,667 +0,0 @@
-;
-;  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-;  Use of this source code is governed by a BSD-style license
-;  that can be found in the LICENSE file in the root of the source
-;  tree. An additional intellectual property rights grant can be found
-;  in the file PATENTS.  All contributing project authors may
-;  be found in the AUTHORS file in the root of the source tree.
-;
-
-%include "third_party/x86inc/x86inc.asm"
-
-SECTION_RODATA
-pw_4:  times 8 dw 4
-pw_8:  times 8 dw 8
-pw_16: times 8 dw 16
-pw_32: times 8 dw 32
-dc_128: times 16 db 128
-pw2_4:  times 8 dw 2
-pw2_8:  times 8 dw 4
-pw2_16:  times 8 dw 8
-pw2_32:  times 8 dw 16
-
-SECTION .text
-
-INIT_MMX sse
-cglobal dc_predictor_4x4, 4, 5, 2, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  movd                  m0, [aboveq]
-  punpckldq             m0, [leftq]
-  psadbw                m0, m1
-  paddw                 m0, [GLOBAL(pw_4)]
-  psraw                 m0, 3
-  pshufw                m0, m0, 0x0
-  packuswb              m0, m0
-  movd      [dstq        ], m0
-  movd      [dstq+strideq], m0
-  lea                 dstq, [dstq+strideq*2]
-  movd      [dstq        ], m0
-  movd      [dstq+strideq], m0
-
-  RESTORE_GOT
-  RET
-
-INIT_MMX sse
-cglobal dc_left_predictor_4x4, 4, 5, 2, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  movd                  m0, [leftq]
-  psadbw                m0, m1
-  paddw                 m0, [GLOBAL(pw2_4)]
-  psraw                 m0, 2
-  pshufw                m0, m0, 0x0
-  packuswb              m0, m0
-  movd      [dstq        ], m0
-  movd      [dstq+strideq], m0
-  lea                 dstq, [dstq+strideq*2]
-  movd      [dstq        ], m0
-  movd      [dstq+strideq], m0
-
-  RESTORE_GOT
-  RET
-
-INIT_MMX sse
-cglobal dc_top_predictor_4x4, 4, 5, 2, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  movd                  m0, [aboveq]
-  psadbw                m0, m1
-  paddw                 m0, [GLOBAL(pw2_4)]
-  psraw                 m0, 2
-  pshufw                m0, m0, 0x0
-  packuswb              m0, m0
-  movd      [dstq        ], m0
-  movd      [dstq+strideq], m0
-  lea                 dstq, [dstq+strideq*2]
-  movd      [dstq        ], m0
-  movd      [dstq+strideq], m0
-
-  RESTORE_GOT
-  RET
-
-INIT_MMX sse
-cglobal dc_predictor_8x8, 4, 5, 3, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  movq                  m0, [aboveq]
-  movq                  m2, [leftq]
-  DEFINE_ARGS dst, stride, stride3
-  lea             stride3q, [strideq*3]
-  psadbw                m0, m1
-  psadbw                m2, m1
-  paddw                 m0, m2
-  paddw                 m0, [GLOBAL(pw_8)]
-  psraw                 m0, 4
-  pshufw                m0, m0, 0x0
-  packuswb              m0, m0
-  movq    [dstq          ], m0
-  movq    [dstq+strideq  ], m0
-  movq    [dstq+strideq*2], m0
-  movq    [dstq+stride3q ], m0
-  lea                 dstq, [dstq+strideq*4]
-  movq    [dstq          ], m0
-  movq    [dstq+strideq  ], m0
-  movq    [dstq+strideq*2], m0
-  movq    [dstq+stride3q ], m0
-
-  RESTORE_GOT
-  RET
-
-INIT_MMX sse
-cglobal dc_top_predictor_8x8, 4, 5, 3, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  movq                  m0, [aboveq]
-  DEFINE_ARGS dst, stride, stride3
-  lea             stride3q, [strideq*3]
-  psadbw                m0, m1
-  paddw                 m0, [GLOBAL(pw2_8)]
-  psraw                 m0, 3
-  pshufw                m0, m0, 0x0
-  packuswb              m0, m0
-  movq    [dstq          ], m0
-  movq    [dstq+strideq  ], m0
-  movq    [dstq+strideq*2], m0
-  movq    [dstq+stride3q ], m0
-  lea                 dstq, [dstq+strideq*4]
-  movq    [dstq          ], m0
-  movq    [dstq+strideq  ], m0
-  movq    [dstq+strideq*2], m0
-  movq    [dstq+stride3q ], m0
-
-  RESTORE_GOT
-  RET
-
-INIT_MMX sse
-cglobal dc_left_predictor_8x8, 4, 5, 3, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  movq                  m0, [leftq]
-  DEFINE_ARGS dst, stride, stride3
-  lea             stride3q, [strideq*3]
-  psadbw                m0, m1
-  paddw                 m0, [GLOBAL(pw2_8)]
-  psraw                 m0, 3
-  pshufw                m0, m0, 0x0
-  packuswb              m0, m0
-  movq    [dstq          ], m0
-  movq    [dstq+strideq  ], m0
-  movq    [dstq+strideq*2], m0
-  movq    [dstq+stride3q ], m0
-  lea                 dstq, [dstq+strideq*4]
-  movq    [dstq          ], m0
-  movq    [dstq+strideq  ], m0
-  movq    [dstq+strideq*2], m0
-  movq    [dstq+stride3q ], m0
-
-  RESTORE_GOT
-  RET
-
-INIT_MMX sse
-cglobal dc_128_predictor_4x4, 4, 5, 3, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  DEFINE_ARGS dst, stride, stride3
-  lea             stride3q, [strideq*3]
-  movd     m0,        [GLOBAL(dc_128)]
-  movd    [dstq          ], m0
-  movd    [dstq+strideq  ], m0
-  movd    [dstq+strideq*2], m0
-  movd    [dstq+stride3q ], m0
-  RESTORE_GOT
-  RET
-
-INIT_MMX sse
-cglobal dc_128_predictor_8x8, 4, 5, 3, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  DEFINE_ARGS dst, stride, stride3
-  lea             stride3q, [strideq*3]
-  movq    m0,        [GLOBAL(dc_128)]
-  movq    [dstq          ], m0
-  movq    [dstq+strideq  ], m0
-  movq    [dstq+strideq*2], m0
-  movq    [dstq+stride3q ], m0
-  lea                 dstq, [dstq+strideq*4]
-  movq    [dstq          ], m0
-  movq    [dstq+strideq  ], m0
-  movq    [dstq+strideq*2], m0
-  movq    [dstq+stride3q ], m0
-  RESTORE_GOT
-  RET
-
-INIT_XMM sse2
-cglobal dc_predictor_16x16, 4, 5, 3, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  mova                  m0, [aboveq]
-  mova                  m2, [leftq]
-  DEFINE_ARGS dst, stride, stride3, lines4
-  lea             stride3q, [strideq*3]
-  mov              lines4d, 4
-  psadbw                m0, m1
-  psadbw                m2, m1
-  paddw                 m0, m2
-  movhlps               m2, m0
-  paddw                 m0, m2
-  paddw                 m0, [GLOBAL(pw_16)]
-  psraw                 m0, 5
-  pshuflw               m0, m0, 0x0
-  punpcklqdq            m0, m0
-  packuswb              m0, m0
-.loop:
-  mova    [dstq          ], m0
-  mova    [dstq+strideq  ], m0
-  mova    [dstq+strideq*2], m0
-  mova    [dstq+stride3q ], m0
-  lea                 dstq, [dstq+strideq*4]
-  dec              lines4d
-  jnz .loop
-
-  RESTORE_GOT
-  REP_RET
-
-
-INIT_XMM sse2
-cglobal dc_top_predictor_16x16, 4, 5, 3, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  pxor                  m2, m2
-  mova                  m0, [aboveq]
-  DEFINE_ARGS dst, stride, stride3, lines4
-  lea             stride3q, [strideq*3]
-  mov              lines4d, 4
-  psadbw                m0, m1
-  psadbw                m2, m1
-  paddw                 m0, m2
-  movhlps               m2, m0
-  paddw                 m0, m2
-  paddw                 m0, [GLOBAL(pw2_16)]
-  psraw                 m0, 4
-  pshuflw               m0, m0, 0x0
-  punpcklqdq            m0, m0
-  packuswb              m0, m0
-.loop:
-  mova    [dstq          ], m0
-  mova    [dstq+strideq  ], m0
-  mova    [dstq+strideq*2], m0
-  mova    [dstq+stride3q ], m0
-  lea                 dstq, [dstq+strideq*4]
-  dec              lines4d
-  jnz .loop
-
-  RESTORE_GOT
-  REP_RET
-
-INIT_XMM sse2
-cglobal dc_left_predictor_16x16, 4, 5, 3, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  pxor                  m2, m2
-  mova                  m0, [leftq]
-  DEFINE_ARGS dst, stride, stride3, lines4
-  lea             stride3q, [strideq*3]
-  mov              lines4d, 4
-  psadbw                m0, m1
-  psadbw                m2, m1
-  paddw                 m0, m2
-  movhlps               m2, m0
-  paddw                 m0, m2
-  paddw                 m0, [GLOBAL(pw2_16)]
-  psraw                 m0, 4
-  pshuflw               m0, m0, 0x0
-  punpcklqdq            m0, m0
-  packuswb              m0, m0
-.loop:
-  mova    [dstq          ], m0
-  mova    [dstq+strideq  ], m0
-  mova    [dstq+strideq*2], m0
-  mova    [dstq+stride3q ], m0
-  lea                 dstq, [dstq+strideq*4]
-  dec              lines4d
-  jnz .loop
-
-  RESTORE_GOT
-  REP_RET
-
-INIT_XMM sse2
-cglobal dc_128_predictor_16x16, 4, 5, 3, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  DEFINE_ARGS dst, stride, stride3, lines4
-  lea             stride3q, [strideq*3]
-  mov              lines4d, 4
-  mova    m0,        [GLOBAL(dc_128)]
-.loop:
-  mova    [dstq          ], m0
-  mova    [dstq+strideq  ], m0
-  mova    [dstq+strideq*2], m0
-  mova    [dstq+stride3q ], m0
-  lea                 dstq, [dstq+strideq*4]
-  dec              lines4d
-  jnz .loop
-  RESTORE_GOT
-  RET
-
-
-INIT_XMM sse2
-cglobal dc_predictor_32x32, 4, 5, 5, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  mova                  m0, [aboveq]
-  mova                  m2, [aboveq+16]
-  mova                  m3, [leftq]
-  mova                  m4, [leftq+16]
-  DEFINE_ARGS dst, stride, stride3, lines4
-  lea             stride3q, [strideq*3]
-  mov              lines4d, 8
-  psadbw                m0, m1
-  psadbw                m2, m1
-  psadbw                m3, m1
-  psadbw                m4, m1
-  paddw                 m0, m2
-  paddw                 m0, m3
-  paddw                 m0, m4
-  movhlps               m2, m0
-  paddw                 m0, m2
-  paddw                 m0, [GLOBAL(pw_32)]
-  psraw                 m0, 6
-  pshuflw               m0, m0, 0x0
-  punpcklqdq            m0, m0
-  packuswb              m0, m0
-.loop:
-  mova [dstq             ], m0
-  mova [dstq          +16], m0
-  mova [dstq+strideq     ], m0
-  mova [dstq+strideq  +16], m0
-  mova [dstq+strideq*2   ], m0
-  mova [dstq+strideq*2+16], m0
-  mova [dstq+stride3q    ], m0
-  mova [dstq+stride3q +16], m0
-  lea                 dstq, [dstq+strideq*4]
-  dec              lines4d
-  jnz .loop
-
-  RESTORE_GOT
-  REP_RET
-
-INIT_XMM sse2
-cglobal dc_top_predictor_32x32, 4, 5, 5, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  mova                  m0, [aboveq]
-  mova                  m2, [aboveq+16]
-  DEFINE_ARGS dst, stride, stride3, lines4
-  lea             stride3q, [strideq*3]
-  mov              lines4d, 8
-  psadbw                m0, m1
-  psadbw                m2, m1
-  paddw                 m0, m2
-  movhlps               m2, m0
-  paddw                 m0, m2
-  paddw                 m0, [GLOBAL(pw2_32)]
-  psraw                 m0, 5
-  pshuflw               m0, m0, 0x0
-  punpcklqdq            m0, m0
-  packuswb              m0, m0
-.loop:
-  mova [dstq             ], m0
-  mova [dstq          +16], m0
-  mova [dstq+strideq     ], m0
-  mova [dstq+strideq  +16], m0
-  mova [dstq+strideq*2   ], m0
-  mova [dstq+strideq*2+16], m0
-  mova [dstq+stride3q    ], m0
-  mova [dstq+stride3q +16], m0
-  lea                 dstq, [dstq+strideq*4]
-  dec              lines4d
-  jnz .loop
-
-  RESTORE_GOT
-  REP_RET
-
-INIT_XMM sse2
-cglobal dc_left_predictor_32x32, 4, 5, 5, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  pxor                  m1, m1
-  mova                  m0, [leftq]
-  mova                  m2, [leftq+16]
-  DEFINE_ARGS dst, stride, stride3, lines4
-  lea             stride3q, [strideq*3]
-  mov              lines4d, 8
-  psadbw                m0, m1
-  psadbw                m2, m1
-  paddw                 m0, m2
-  movhlps               m2, m0
-  paddw                 m0, m2
-  paddw                 m0, [GLOBAL(pw2_32)]
-  psraw                 m0, 5
-  pshuflw               m0, m0, 0x0
-  punpcklqdq            m0, m0
-  packuswb              m0, m0
-.loop:
-  mova [dstq             ], m0
-  mova [dstq          +16], m0
-  mova [dstq+strideq     ], m0
-  mova [dstq+strideq  +16], m0
-  mova [dstq+strideq*2   ], m0
-  mova [dstq+strideq*2+16], m0
-  mova [dstq+stride3q    ], m0
-  mova [dstq+stride3q +16], m0
-  lea                 dstq, [dstq+strideq*4]
-  dec              lines4d
-  jnz .loop
-
-  RESTORE_GOT
-  REP_RET
-
-INIT_XMM sse2
-cglobal dc_128_predictor_32x32, 4, 5, 3, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-
-  DEFINE_ARGS dst, stride, stride3, lines4
-  lea             stride3q, [strideq*3]
-  mov              lines4d, 8
-  mova    m0,        [GLOBAL(dc_128)]
-.loop:
-  mova [dstq             ], m0
-  mova [dstq          +16], m0
-  mova [dstq+strideq     ], m0
-  mova [dstq+strideq  +16], m0
-  mova [dstq+strideq*2   ], m0
-  mova [dstq+strideq*2+16], m0
-  mova [dstq+stride3q    ], m0
-  mova [dstq+stride3q +16], m0
-  lea                 dstq, [dstq+strideq*4]
-  dec              lines4d
-  jnz .loop
-  RESTORE_GOT
-  RET
-
-INIT_MMX sse
-cglobal v_predictor_4x4, 3, 3, 1, dst, stride, above
-  movd                  m0, [aboveq]
-  movd      [dstq        ], m0
-  movd      [dstq+strideq], m0
-  lea                 dstq, [dstq+strideq*2]
-  movd      [dstq        ], m0
-  movd      [dstq+strideq], m0
-  RET
-
-INIT_MMX sse
-cglobal v_predictor_8x8, 3, 3, 1, dst, stride, above
-  movq                  m0, [aboveq]
-  DEFINE_ARGS dst, stride, stride3
-  lea             stride3q, [strideq*3]
-  movq    [dstq          ], m0
-  movq    [dstq+strideq  ], m0
-  movq    [dstq+strideq*2], m0
-  movq    [dstq+stride3q ], m0
-  lea                 dstq, [dstq+strideq*4]
-  movq    [dstq          ], m0
-  movq    [dstq+strideq  ], m0
-  movq    [dstq+strideq*2], m0
-  movq    [dstq+stride3q ], m0
-  RET
-
-INIT_XMM sse2
-cglobal v_predictor_16x16, 3, 4, 1, dst, stride, above
-  mova                  m0, [aboveq]
-  DEFINE_ARGS dst, stride, stride3, nlines4
-  lea             stride3q, [strideq*3]
-  mov              nlines4d, 4
-.loop:
-  mova    [dstq          ], m0
-  mova    [dstq+strideq  ], m0
-  mova    [dstq+strideq*2], m0
-  mova    [dstq+stride3q ], m0
-  lea                 dstq, [dstq+strideq*4]
-  dec             nlines4d
-  jnz .loop
-  REP_RET
-
-INIT_XMM sse2
-cglobal v_predictor_32x32, 3, 4, 2, dst, stride, above
-  mova                  m0, [aboveq]
-  mova                  m1, [aboveq+16]
-  DEFINE_ARGS dst, stride, stride3, nlines4
-  lea             stride3q, [strideq*3]
-  mov              nlines4d, 8
-.loop:
-  mova [dstq             ], m0
-  mova [dstq          +16], m1
-  mova [dstq+strideq     ], m0
-  mova [dstq+strideq  +16], m1
-  mova [dstq+strideq*2   ], m0
-  mova [dstq+strideq*2+16], m1
-  mova [dstq+stride3q    ], m0
-  mova [dstq+stride3q +16], m1
-  lea                 dstq, [dstq+strideq*4]
-  dec             nlines4d
-  jnz .loop
-  REP_RET
-
-INIT_MMX sse
-cglobal tm_predictor_4x4, 4, 4, 4, dst, stride, above, left
-  pxor                  m1, m1
-  movd                  m2, [aboveq-1]
-  movd                  m0, [aboveq]
-  punpcklbw             m2, m1
-  punpcklbw             m0, m1
-  pshufw                m2, m2, 0x0
-  DEFINE_ARGS dst, stride, line, left
-  mov                lineq, -2
-  add                leftq, 4
-  psubw                 m0, m2
-.loop:
-  movd                  m2, [leftq+lineq*2]
-  movd                  m3, [leftq+lineq*2+1]
-  punpcklbw             m2, m1
-  punpcklbw             m3, m1
-  pshufw                m2, m2, 0x0
-  pshufw                m3, m3, 0x0
-  paddw                 m2, m0
-  paddw                 m3, m0
-  packuswb              m2, m2
-  packuswb              m3, m3
-  movd      [dstq        ], m2
-  movd      [dstq+strideq], m3
-  lea                 dstq, [dstq+strideq*2]
-  inc                lineq
-  jnz .loop
-  REP_RET
-
-INIT_XMM sse2
-cglobal tm_predictor_8x8, 4, 4, 4, dst, stride, above, left
-  pxor                  m1, m1
-  movd                  m2, [aboveq-1]
-  movq                  m0, [aboveq]
-  punpcklbw             m2, m1
-  punpcklbw             m0, m1
-  pshuflw               m2, m2, 0x0
-  DEFINE_ARGS dst, stride, line, left
-  mov                lineq, -4
-  punpcklqdq            m2, m2
-  add                leftq, 8
-  psubw                 m0, m2
-.loop:
-  movd                  m2, [leftq+lineq*2]
-  movd                  m3, [leftq+lineq*2+1]
-  punpcklbw             m2, m1
-  punpcklbw             m3, m1
-  pshuflw               m2, m2, 0x0
-  pshuflw               m3, m3, 0x0
-  punpcklqdq            m2, m2
-  punpcklqdq            m3, m3
-  paddw                 m2, m0
-  paddw                 m3, m0
-  packuswb              m2, m3
-  movq      [dstq        ], m2
-  movhps    [dstq+strideq], m2
-  lea                 dstq, [dstq+strideq*2]
-  inc                lineq
-  jnz .loop
-  REP_RET
-
-INIT_XMM sse2
-cglobal tm_predictor_16x16, 4, 4, 7, dst, stride, above, left
-  pxor                  m1, m1
-  movd                  m2, [aboveq-1]
-  mova                  m0, [aboveq]
-  punpcklbw             m2, m1
-  punpckhbw             m4, m0, m1
-  punpcklbw             m0, m1
-  pshuflw               m2, m2, 0x0
-  DEFINE_ARGS dst, stride, line, left
-  mov                lineq, -8
-  punpcklqdq            m2, m2
-  add                leftq, 16
-  psubw                 m0, m2
-  psubw                 m4, m2
-.loop:
-  movd                  m2, [leftq+lineq*2]
-  movd                  m3, [leftq+lineq*2+1]
-  punpcklbw             m2, m1
-  punpcklbw             m3, m1
-  pshuflw               m2, m2, 0x0
-  pshuflw               m3, m3, 0x0
-  punpcklqdq            m2, m2
-  punpcklqdq            m3, m3
-  paddw                 m5, m2, m0
-  paddw                 m6, m3, m0
-  paddw                 m2, m4
-  paddw                 m3, m4
-  packuswb              m5, m2
-  packuswb              m6, m3
-  mova      [dstq        ], m5
-  mova      [dstq+strideq], m6
-  lea                 dstq, [dstq+strideq*2]
-  inc                lineq
-  jnz .loop
-  REP_RET
-
-%if ARCH_X86_64
-INIT_XMM sse2
-cglobal tm_predictor_32x32, 4, 4, 10, dst, stride, above, left
-  pxor                  m1, m1
-  movd                  m2, [aboveq-1]
-  mova                  m0, [aboveq]
-  mova                  m4, [aboveq+16]
-  punpcklbw             m2, m1
-  punpckhbw             m3, m0, m1
-  punpckhbw             m5, m4, m1
-  punpcklbw             m0, m1
-  punpcklbw             m4, m1
-  pshuflw               m2, m2, 0x0
-  DEFINE_ARGS dst, stride, line, left
-  mov                lineq, -16
-  punpcklqdq            m2, m2
-  add                leftq, 32
-  psubw                 m0, m2
-  psubw                 m3, m2
-  psubw                 m4, m2
-  psubw                 m5, m2
-.loop:
-  movd                  m2, [leftq+lineq*2]
-  movd                  m6, [leftq+lineq*2+1]
-  punpcklbw             m2, m1
-  punpcklbw             m6, m1
-  pshuflw               m2, m2, 0x0
-  pshuflw               m6, m6, 0x0
-  punpcklqdq            m2, m2
-  punpcklqdq            m6, m6
-  paddw                 m7, m2, m0
-  paddw                 m8, m2, m3
-  paddw                 m9, m2, m4
-  paddw                 m2, m5
-  packuswb              m7, m8
-  packuswb              m9, m2
-  paddw                 m2, m6, m0
-  paddw                 m8, m6, m3
-  mova   [dstq           ], m7
-  paddw                 m7, m6, m4
-  paddw                 m6, m5
-  mova   [dstq        +16], m9
-  packuswb              m2, m8
-  packuswb              m7, m6
-  mova   [dstq+strideq   ], m2
-  mova   [dstq+strideq+16], m7
-  lea                 dstq, [dstq+strideq*2]
-  inc                lineq
-  jnz .loop
-  REP_RET
-%endif
diff --git a/vp9/common/x86/vp9_intrapred_ssse3.asm b/vp9/common/x86/vp9_intrapred_ssse3.asm
deleted file mode 100644
index 88df9b2d1..000000000
--- a/vp9/common/x86/vp9_intrapred_ssse3.asm
+++ /dev/null
@@ -1,1036 +0,0 @@
-;
-;  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-;  Use of this source code is governed by a BSD-style license
-;  that can be found in the LICENSE file in the root of the source
-;  tree. An additional intellectual property rights grant can be found
-;  in the file PATENTS.  All contributing project authors may
-;  be found in the AUTHORS file in the root of the source tree.
-;
-
-%include "third_party/x86inc/x86inc.asm"
-
-SECTION_RODATA
-
-pb_1: times 16 db 1
-sh_b01234577: db 0, 1, 2, 3, 4, 5, 7, 7, 0, 0, 0, 0, 0, 0, 0, 0
-sh_b12345677: db 1, 2, 3, 4, 5, 6, 7, 7, 0, 0, 0, 0, 0, 0, 0, 0
-sh_b23456777: db 2, 3, 4, 5, 6, 7, 7, 7, 0, 0, 0, 0, 0, 0, 0, 0
-sh_b0123456777777777: db 0, 1, 2, 3, 4, 5, 6, 7, 7, 7, 7, 7, 7, 7, 7, 7
-sh_b1234567777777777: db 1, 2, 3, 4, 5, 6, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7
-sh_b2345677777777777: db 2, 3, 4, 5, 6, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7
-sh_b123456789abcdeff: db 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 15
-sh_b23456789abcdefff: db 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 15, 15
-sh_b32104567: db 3, 2, 1, 0, 4, 5, 6, 7, 0, 0, 0, 0, 0, 0, 0, 0
-sh_b8091a2b345: db 8, 0, 9, 1, 10, 2, 11, 3, 4, 5, 0, 0, 0, 0, 0, 0
-sh_b76543210: db 7, 6, 5, 4, 3, 2, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0
-sh_b65432108: db 6, 5, 4, 3, 2, 1, 0, 8, 0, 0, 0, 0, 0, 0, 0, 0
-sh_b54321089: db 5, 4, 3, 2, 1, 0, 8, 9, 0, 0, 0, 0, 0, 0, 0, 0
-sh_b89abcdef: db 8, 9, 10, 11, 12, 13, 14, 15, 0, 0, 0, 0, 0, 0, 0, 0
-sh_bfedcba9876543210: db 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0
-sh_b1233: db 1, 2, 3, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
-sh_b2333: db 2, 3, 3, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
-
-SECTION .text
-
-INIT_MMX ssse3
-cglobal h_predictor_4x4, 2, 4, 3, dst, stride, line, left
-  movifnidn          leftq, leftmp
-  add                leftq, 4
-  mov                lineq, -2
-  pxor                  m0, m0
-.loop:
-  movd                  m1, [leftq+lineq*2  ]
-  movd                  m2, [leftq+lineq*2+1]
-  pshufb                m1, m0
-  pshufb                m2, m0
-  movd      [dstq        ], m1
-  movd      [dstq+strideq], m2
-  lea                 dstq, [dstq+strideq*2]
-  inc                lineq
-  jnz .loop
-  REP_RET
-
-INIT_MMX ssse3
-cglobal h_predictor_8x8, 2, 4, 3, dst, stride, line, left
-  movifnidn          leftq, leftmp
-  add                leftq, 8
-  mov                lineq, -4
-  pxor                  m0, m0
-.loop:
-  movd                  m1, [leftq+lineq*2  ]
-  movd                  m2, [leftq+lineq*2+1]
-  pshufb                m1, m0
-  pshufb                m2, m0
-  movq      [dstq        ], m1
-  movq      [dstq+strideq], m2
-  lea                 dstq, [dstq+strideq*2]
-  inc                lineq
-  jnz .loop
-  REP_RET
-
-INIT_XMM ssse3
-cglobal h_predictor_16x16, 2, 4, 3, dst, stride, line, left
-  movifnidn          leftq, leftmp
-  add                leftq, 16
-  mov                lineq, -8
-  pxor                  m0, m0
-.loop:
-  movd                  m1, [leftq+lineq*2  ]
-  movd                  m2, [leftq+lineq*2+1]
-  pshufb                m1, m0
-  pshufb                m2, m0
-  mova      [dstq        ], m1
-  mova      [dstq+strideq], m2
-  lea                 dstq, [dstq+strideq*2]
-  inc                lineq
-  jnz .loop
-  REP_RET
-
-INIT_XMM ssse3
-cglobal h_predictor_32x32, 2, 4, 3, dst, stride, line, left
-  movifnidn          leftq, leftmp
-  add                leftq, 32
-  mov                lineq, -16
-  pxor                  m0, m0
-.loop:
-  movd                  m1, [leftq+lineq*2  ]
-  movd                  m2, [leftq+lineq*2+1]
-  pshufb                m1, m0
-  pshufb                m2, m0
-  mova   [dstq           ], m1
-  mova   [dstq        +16], m1
-  mova   [dstq+strideq   ], m2
-  mova   [dstq+strideq+16], m2
-  lea                 dstq, [dstq+strideq*2]
-  inc                lineq
-  jnz .loop
-  REP_RET
-
-INIT_MMX ssse3
-cglobal d45_predictor_4x4, 3, 4, 4, dst, stride, above, goffset
-  GET_GOT     goffsetq
-
-  movq                m0, [aboveq]
-  pshufb              m2, m0, [GLOBAL(sh_b23456777)]
-  pshufb              m1, m0, [GLOBAL(sh_b01234577)]
-  pshufb              m0, [GLOBAL(sh_b12345677)]
-  pavgb               m3, m2, m1
-  pxor                m2, m1
-  pand                m2, [GLOBAL(pb_1)]
-  psubb               m3, m2
-  pavgb               m0, m3
-
-  ; store 4 lines
-  movd    [dstq        ], m0
-  psrlq               m0, 8
-  movd    [dstq+strideq], m0
-  lea               dstq, [dstq+strideq*2]
-  psrlq               m0, 8
-  movd    [dstq        ], m0
-  psrlq               m0, 8
-  movd    [dstq+strideq], m0
-
-  RESTORE_GOT
-  RET
-
-INIT_MMX ssse3
-cglobal d45_predictor_8x8, 3, 4, 4, dst, stride, above, goffset
-  GET_GOT     goffsetq
-
-  movq                m0, [aboveq]
-  mova                m1, [GLOBAL(sh_b12345677)]
-  DEFINE_ARGS dst, stride, stride3
-  lea           stride3q, [strideq*3]
-  pshufb              m2, m0, [GLOBAL(sh_b23456777)]
-  pavgb               m3, m2, m0
-  pxor                m2, m0
-  pshufb              m0, m1
-  pand                m2, [GLOBAL(pb_1)]
-  psubb               m3, m2
-  pavgb               m0, m3
-
-  ; store 4 lines
-  movq  [dstq          ], m0
-  pshufb              m0, m1
-  movq  [dstq+strideq  ], m0
-  pshufb              m0, m1
-  movq  [dstq+strideq*2], m0
-  pshufb              m0, m1
-  movq  [dstq+stride3q ], m0
-  pshufb              m0, m1
-  lea               dstq, [dstq+strideq*4]
-
-  ; store next 4 lines
-  movq  [dstq          ], m0
-  pshufb              m0, m1
-  movq  [dstq+strideq  ], m0
-  pshufb              m0, m1
-  movq  [dstq+strideq*2], m0
-  pshufb              m0, m1
-  movq  [dstq+stride3q ], m0
-
-  RESTORE_GOT
-  RET
-
-INIT_XMM ssse3
-cglobal d45_predictor_16x16, 3, 6, 4, dst, stride, above, dst8, line, goffset
-  GET_GOT     goffsetq
-
-  mova                   m0, [aboveq]
-  DEFINE_ARGS dst, stride, stride3, dst8, line
-  lea              stride3q, [strideq*3]
-  lea                 dst8q, [dstq+strideq*8]
-  mova                   m1, [GLOBAL(sh_b123456789abcdeff)]
-  pshufb                 m2, m0, [GLOBAL(sh_b23456789abcdefff)]
-  pavgb                  m3, m2, m0
-  pxor                   m2, m0
-  pshufb                 m0, m1
-  pand                   m2, [GLOBAL(pb_1)]
-  psubb                  m3, m2
-  pavgb                  m0, m3
-
-  ; first 4 lines and first half of 3rd 4 lines
-  mov                 lined, 2
-.loop:
-  mova   [dstq            ], m0
-  movhps [dst8q           ], m0
-  pshufb                 m0, m1
-  mova   [dstq +strideq   ], m0
-  movhps [dst8q+strideq   ], m0
-  pshufb                 m0, m1
-  mova   [dstq +strideq*2 ], m0
-  movhps [dst8q+strideq*2 ], m0
-  pshufb                 m0, m1
-  mova   [dstq +stride3q  ], m0
-  movhps [dst8q+stride3q  ], m0
-  pshufb                 m0, m1
-  lea                  dstq, [dstq +strideq*4]
-  lea                 dst8q, [dst8q+strideq*4]
-  dec                 lined
-  jnz .loop
-
-  ; bottom-right 8x8 block
-  movhps [dstq          +8], m0
-  movhps [dstq+strideq  +8], m0
-  movhps [dstq+strideq*2+8], m0
-  movhps [dstq+stride3q +8], m0
-  lea                  dstq, [dstq+strideq*4]
-  movhps [dstq          +8], m0
-  movhps [dstq+strideq  +8], m0
-  movhps [dstq+strideq*2+8], m0
-  movhps [dstq+stride3q +8], m0
-
-  RESTORE_GOT
-  RET
-
-INIT_XMM ssse3
-cglobal d45_predictor_32x32, 3, 6, 7, dst, stride, above, dst16, line, goffset
-  GET_GOT     goffsetq
-
-  mova                   m0, [aboveq]
-  mova                   m4, [aboveq+16]
-  DEFINE_ARGS dst, stride, stride3, dst16, line
-  lea              stride3q, [strideq*3]
-  lea                dst16q, [dstq  +strideq*8]
-  lea                dst16q, [dst16q+strideq*8]
-  mova                   m1, [GLOBAL(sh_b123456789abcdeff)]
-  pshufb                 m2, m4, [GLOBAL(sh_b23456789abcdefff)]
-  pavgb                  m3, m2, m4
-  pxor                   m2, m4
-  palignr                m5, m4, m0, 1
-  palignr                m6, m4, m0, 2
-  pshufb                 m4, m1
-  pand                   m2, [GLOBAL(pb_1)]
-  psubb                  m3, m2
-  pavgb                  m4, m3
-  pavgb                  m3, m0, m6
-  pxor                   m0, m6
-  pand                   m0, [GLOBAL(pb_1)]
-  psubb                  m3, m0
-  pavgb                  m5, m3
-
-  ; write 4x4 lines (and the first half of the second 4x4 lines)
-  mov                  lined, 4
-.loop:
-  mova [dstq               ], m5
-  mova [dstq            +16], m4
-  mova [dst16q             ], m4
-  palignr                 m3, m4, m5, 1
-  pshufb                  m4, m1
-  mova [dstq  +strideq     ], m3
-  mova [dstq  +strideq  +16], m4
-  mova [dst16q+strideq     ], m4
-  palignr                 m5, m4, m3, 1
-  pshufb                  m4, m1
-  mova [dstq  +strideq*2   ], m5
-  mova [dstq  +strideq*2+16], m4
-  mova [dst16q+strideq*2   ], m4
-  palignr                 m3, m4, m5, 1
-  pshufb                  m4, m1
-  mova [dstq  +stride3q    ], m3
-  mova [dstq  +stride3q +16], m4
-  mova [dst16q+stride3q    ], m4
-  palignr                 m5, m4, m3, 1
-  pshufb                  m4, m1
-  lea                  dstq, [dstq  +strideq*4]
-  lea                dst16q, [dst16q+strideq*4]
-  dec                 lined
-  jnz .loop
-
-  ; write second half of second 4x4 lines
-  mova [dstq            +16], m4
-  mova [dstq  +strideq  +16], m4
-  mova [dstq  +strideq*2+16], m4
-  mova [dstq  +stride3q +16], m4
-  lea                  dstq, [dstq  +strideq*4]
-  mova [dstq            +16], m4
-  mova [dstq  +strideq  +16], m4
-  mova [dstq  +strideq*2+16], m4
-  mova [dstq  +stride3q +16], m4
-  lea                  dstq, [dstq  +strideq*4]
-  mova [dstq            +16], m4
-  mova [dstq  +strideq  +16], m4
-  mova [dstq  +strideq*2+16], m4
-  mova [dstq  +stride3q +16], m4
-  lea                  dstq, [dstq  +strideq*4]
-  mova [dstq            +16], m4
-  mova [dstq  +strideq  +16], m4
-  mova [dstq  +strideq*2+16], m4
-  mova [dstq  +stride3q +16], m4
-
-  RESTORE_GOT
-  RET
-
-; ------------------------------------------
-; input: x, y, z, result
-;
-; trick from pascal
-; (x+2y+z+2)>>2 can be calculated as:
-; result = avg(x,z)
-; result -= xor(x,z) & 1
-; result = avg(result,y)
-; ------------------------------------------
-%macro X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 4
-  pavgb               %4, %1, %3
-  pxor                %3, %1
-  pand                %3, [GLOBAL(pb_1)]
-  psubb               %4, %3
-  pavgb               %4, %2
-%endmacro
-
-INIT_XMM ssse3
-cglobal d63_predictor_4x4, 3, 4, 5, dst, stride, above, goffset
-  GET_GOT     goffsetq
-
-  movq                m3, [aboveq]
-  pshufb              m1, m3, [GLOBAL(sh_b23456777)]
-  pshufb              m2, m3, [GLOBAL(sh_b12345677)]
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m3, m2, m1, m4
-  pavgb               m3, m2
-
-  ; store 4 lines
-  movd    [dstq        ], m3
-  movd    [dstq+strideq], m4
-  lea               dstq, [dstq+strideq*2]
-  psrldq              m3, 1
-  psrldq              m4, 1
-  movd    [dstq        ], m3
-  movd    [dstq+strideq], m4
-  RESTORE_GOT
-  RET
-
-INIT_XMM ssse3
-cglobal d63_predictor_8x8, 3, 4, 5, dst, stride, above, goffset
-  GET_GOT     goffsetq
-
-  movq                m3, [aboveq]
-  DEFINE_ARGS dst, stride, stride3
-  lea           stride3q, [strideq*3]
-  pshufb              m1, m3, [GLOBAL(sh_b2345677777777777)]
-  pshufb              m0, m3, [GLOBAL(sh_b0123456777777777)]
-  pshufb              m2, m3, [GLOBAL(sh_b1234567777777777)]
-  pshufb              m3, [GLOBAL(sh_b0123456777777777)]
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m0, m2, m1, m4
-  pavgb               m3, m2
-
-  ; store 4 lines
-  movq    [dstq        ], m3
-  movq    [dstq+strideq], m4
-  psrldq              m3, 1
-  psrldq              m4, 1
-  movq  [dstq+strideq*2], m3
-  movq  [dstq+stride3q ], m4
-  lea               dstq, [dstq+strideq*4]
-  psrldq              m3, 1
-  psrldq              m4, 1
-
-  ; store 4 lines
-  movq    [dstq        ], m3
-  movq    [dstq+strideq], m4
-  psrldq              m3, 1
-  psrldq              m4, 1
-  movq  [dstq+strideq*2], m3
-  movq  [dstq+stride3q ], m4
-  RESTORE_GOT
-  RET
-
-INIT_XMM ssse3
-cglobal d63_predictor_16x16, 3, 5, 5, dst, stride, above, line, goffset
-  GET_GOT     goffsetq
-
-  mova                m0, [aboveq]
-  DEFINE_ARGS dst, stride, stride3, line
-  lea           stride3q, [strideq*3]
-  mova                m1, [GLOBAL(sh_b123456789abcdeff)]
-  pshufb              m2, m0, [GLOBAL(sh_b23456789abcdefff)]
-  pshufb              m3, m0, m1
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m0, m3, m2, m4
-  pavgb               m0, m3
-
-  mov              lined, 4
-.loop:
-  mova  [dstq          ], m0
-  mova  [dstq+strideq  ], m4
-  pshufb              m0, m1
-  pshufb              m4, m1
-  mova  [dstq+strideq*2], m0
-  mova  [dstq+stride3q ], m4
-  pshufb              m0, m1
-  pshufb              m4, m1
-  lea               dstq, [dstq+strideq*4]
-  dec              lined
-  jnz .loop
-  RESTORE_GOT
-  REP_RET
-
-INIT_XMM ssse3
-cglobal d63_predictor_32x32, 3, 5, 8, dst, stride, above, line, goffset
-  GET_GOT     goffsetq
-
-  mova                   m0, [aboveq]
-  mova                   m7, [aboveq+16]
-  DEFINE_ARGS dst, stride, stride3, line
-  mova                   m1, [GLOBAL(sh_b123456789abcdeff)]
-  lea              stride3q, [strideq*3]
-  pshufb                 m2, m7, [GLOBAL(sh_b23456789abcdefff)]
-  pshufb                 m3, m7, m1
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m7, m3, m2, m4
-  palignr                m6, m7, m0, 1
-  palignr                m5, m7, m0, 2
-  pavgb                  m7, m3
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m0, m6, m5, m2
-  pavgb                  m0, m6
-
-  mov                 lined, 8
-.loop:
-  mova  [dstq             ], m0
-  mova  [dstq          +16], m7
-  mova  [dstq+strideq     ], m2
-  mova  [dstq+strideq  +16], m4
-  palignr                m3, m7, m0, 1
-  palignr                m5, m4, m2, 1
-  pshufb                 m7, m1
-  pshufb                 m4, m1
-
-  mova  [dstq+strideq*2   ], m3
-  mova  [dstq+strideq*2+16], m7
-  mova  [dstq+stride3q    ], m5
-  mova  [dstq+stride3q +16], m4
-  palignr                m0, m7, m3, 1
-  palignr                m2, m4, m5, 1
-  pshufb                 m7, m1
-  pshufb                 m4, m1
-  lea                  dstq, [dstq+strideq*4]
-  dec                 lined
-  jnz .loop
-  RESTORE_GOT
-  REP_RET
-
-INIT_XMM ssse3
-cglobal d153_predictor_4x4, 4, 5, 4, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-  movd                m0, [leftq]               ; l1, l2, l3, l4
-  movd                m1, [aboveq-1]            ; tl, t1, t2, t3
-  punpckldq           m0, m1                    ; l1, l2, l3, l4, tl, t1, t2, t3
-  pshufb              m0, [GLOBAL(sh_b32104567)]; l4, l3, l2, l1, tl, t1, t2, t3
-  psrldq              m1, m0, 1                 ; l3, l2, l1, tl, t1, t2, t3
-  psrldq              m2, m0, 2                 ; l2, l1, tl, t1, t2, t3
-  ; comments below are for a predictor like this
-  ; A1 B1 C1 D1
-  ; A2 B2 A1 B1
-  ; A3 B3 A2 B2
-  ; A4 B4 A3 B3
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m0, m1, m2, m3  ; 3-tap avg B4 B3 B2 B1 C1 D1
-  pavgb               m1, m0                    ; 2-tap avg A4 A3 A2 A1
-
-  punpcklqdq          m3, m1                    ; B4 B3 B2 B1 C1 D1 x x A4 A3 A2 A1 ..
-
-  DEFINE_ARGS dst, stride, stride3
-  lea           stride3q, [strideq*3]
-  pshufb              m3, [GLOBAL(sh_b8091a2b345)] ; A4 B4 A3 B3 A2 B2 A1 B1 C1 D1 ..
-  movd  [dstq+stride3q ], m3
-  psrldq              m3, 2                     ; A3 B3 A2 B2 A1 B1 C1 D1 ..
-  movd  [dstq+strideq*2], m3
-  psrldq              m3, 2                     ; A2 B2 A1 B1 C1 D1 ..
-  movd  [dstq+strideq  ], m3
-  psrldq              m3, 2                     ; A1 B1 C1 D1 ..
-  movd  [dstq          ], m3
-  RESTORE_GOT
-  RET
-
-INIT_XMM ssse3
-cglobal d153_predictor_8x8, 4, 5, 8, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-  movq                m0, [leftq]                     ; [0- 7] l1-8 [byte]
-  movhps              m0, [aboveq-1]                  ; [8-15] tl, t1-7 [byte]
-  pshufb              m1, m0, [GLOBAL(sh_b76543210)]  ; l8-1 [word]
-  pshufb              m2, m0, [GLOBAL(sh_b65432108)]  ; l7-1,tl [word]
-  pshufb              m3, m0, [GLOBAL(sh_b54321089)]  ; l6-1,tl,t1 [word]
-  pshufb              m0, [GLOBAL(sh_b89abcdef)]      ; tl,t1-7 [word]
-  psrldq              m4, m0, 1                       ; t1-7 [word]
-  psrldq              m5, m0, 2                       ; t2-7 [word]
-  ; comments below are for a predictor like this
-  ; A1 B1 C1 D1 E1 F1 G1 H1
-  ; A2 B2 A1 B1 C1 D1 E1 F1
-  ; A3 B3 A2 B2 A1 B1 C1 D1
-  ; A4 B4 A3 B3 A2 B2 A1 B1
-  ; A5 B5 A4 B4 A3 B3 A2 B2
-  ; A6 B6 A5 B5 A4 B4 A3 B3
-  ; A7 B7 A6 B6 A5 B5 A4 B4
-  ; A8 B8 A7 B7 A6 B6 A5 B5
-  pavgb               m6, m1, m2                ; 2-tap avg A8-A1
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m0, m4, m5, m7  ; 3-tap avg C-H1
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m1, m2, m3, m0  ; 3-tap avg B8-1
-
-  punpcklbw           m6, m0                    ; A-B8, A-B7 ... A-B2, A-B1
-
-  DEFINE_ARGS dst, stride, stride3
-  lea           stride3q, [strideq*3]
-
-  movhps [dstq+stride3q], m6                    ; A-B4, A-B3, A-B2, A-B1
-  palignr             m0, m7, m6, 10            ; A-B3, A-B2, A-B1, C-H1
-  movq  [dstq+strideq*2], m0
-  psrldq              m0, 2                     ; A-B2, A-B1, C-H1
-  movq  [dstq+strideq  ], m0
-  psrldq              m0, 2                     ; A-H1
-  movq  [dstq          ], m0
-  lea               dstq, [dstq+strideq*4]
-  movq  [dstq+stride3q ], m6                    ; A-B8, A-B7, A-B6, A-B5
-  psrldq              m6, 2                     ; A-B7, A-B6, A-B5, A-B4
-  movq  [dstq+strideq*2], m6
-  psrldq              m6, 2                     ; A-B6, A-B5, A-B4, A-B3
-  movq  [dstq+strideq  ], m6
-  psrldq              m6, 2                     ; A-B5, A-B4, A-B3, A-B2
-  movq  [dstq          ], m6
-  RESTORE_GOT
-  RET
-
-INIT_XMM ssse3
-cglobal d153_predictor_16x16, 4, 5, 8, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-  mova                m0, [leftq]
-  movu                m7, [aboveq-1]
-  ; comments below are for a predictor like this
-  ; A1 B1 C1 D1 E1 F1 G1 H1 I1 J1 K1 L1 M1 N1 O1 P1
-  ; A2 B2 A1 B1 C1 D1 E1 F1 G1 H1 I1 J1 K1 L1 M1 N1
-  ; A3 B3 A2 B2 A1 B1 C1 D1 E1 F1 G1 H1 I1 J1 K1 L1
-  ; A4 B4 A3 B3 A2 B2 A1 B1 C1 D1 E1 F1 G1 H1 I1 J1
-  ; A5 B5 A4 B4 A3 B3 A2 B2 A1 B1 C1 D1 E1 F1 G1 H1
-  ; A6 B6 A5 B5 A4 B4 A3 B3 A2 B2 A1 B1 C1 D1 E1 F1
-  ; A7 B7 A6 B6 A5 B5 A4 B4 A3 B3 A2 B2 A1 B1 C1 D1
-  ; A8 B8 A7 B7 A6 B6 A5 B5 A4 B4 A3 B3 A2 B2 A1 B1
-  ; A9 B9 A8 B8 A7 B7 A6 B6 A5 B5 A4 B4 A3 B3 A2 B2
-  ; Aa Ba A9 B9 A8 B8 A7 B7 A6 B6 A5 B5 A4 B4 A3 B3
-  ; Ab Bb Aa Ba A9 B9 A8 B8 A7 B7 A6 B6 A5 B5 A4 B4
-  ; Ac Bc Ab Bb Aa Ba A9 B9 A8 B8 A7 B7 A6 B6 A5 B5
-  ; Ad Bd Ac Bc Ab Bb Aa Ba A9 B9 A8 B8 A7 B7 A6 B6
-  ; Ae Be Ad Bd Ac Bc Ab Bb Aa Ba A9 B9 A8 B8 A7 B7
-  ; Af Bf Ae Be Ad Bd Ac Bc Ab Bb Aa Ba A9 B9 A8 B8
-  ; Ag Bg Af Bf Ae Be Ad Bd Ac Bc Ab Bb Aa Ba A9 B9
-  pshufb              m6, m7, [GLOBAL(sh_bfedcba9876543210)]
-  palignr             m5, m0, m6, 15
-  palignr             m3, m0, m6, 14
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m0, m5, m3, m4          ; 3-tap avg B3-Bg
-  pshufb              m1, m0, [GLOBAL(sh_b123456789abcdeff)]
-  pavgb               m5, m0                            ; A1 - Ag
-
-  punpcklbw           m0, m4, m5                        ; A-B8 ... A-B1
-  punpckhbw           m4, m5                            ; A-B9 ... A-Bg
-
-  pshufb              m3, m7, [GLOBAL(sh_b123456789abcdeff)]
-  pshufb              m5, m7, [GLOBAL(sh_b23456789abcdefff)]
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m7, m3, m5, m1          ; 3-tap avg C1-P1
-
-  pshufb              m6, m0, [GLOBAL(sh_bfedcba9876543210)]
-  DEFINE_ARGS dst, stride, stride3
-  lea           stride3q, [strideq*3]
-  palignr             m2, m1, m6, 14
-  mova  [dstq          ], m2
-  palignr             m2, m1, m6, 12
-  mova  [dstq+strideq  ], m2
-  palignr             m2, m1, m6, 10
-  mova  [dstq+strideq*2], m2
-  palignr             m2, m1, m6, 8
-  mova  [dstq+stride3q ], m2
-  lea               dstq, [dstq+strideq*4]
-  palignr             m2, m1, m6, 6
-  mova  [dstq          ], m2
-  palignr             m2, m1, m6, 4
-  mova  [dstq+strideq  ], m2
-  palignr             m2, m1, m6, 2
-  mova  [dstq+strideq*2], m2
-  pshufb              m4, [GLOBAL(sh_bfedcba9876543210)]
-  mova  [dstq+stride3q ], m6
-  lea               dstq, [dstq+strideq*4]
-
-  palignr             m2, m6, m4, 14
-  mova  [dstq          ], m2
-  palignr             m2, m6, m4, 12
-  mova  [dstq+strideq  ], m2
-  palignr             m2, m6, m4, 10
-  mova  [dstq+strideq*2], m2
-  palignr             m2, m6, m4, 8
-  mova  [dstq+stride3q ], m2
-  lea               dstq, [dstq+strideq*4]
-  palignr             m2, m6, m4, 6
-  mova  [dstq          ], m2
-  palignr             m2, m6, m4, 4
-  mova  [dstq+strideq  ], m2
-  palignr             m2, m6, m4, 2
-  mova  [dstq+strideq*2], m2
-  mova  [dstq+stride3q ], m4
-  RESTORE_GOT
-  RET
-
-INIT_XMM ssse3
-cglobal d153_predictor_32x32, 4, 5, 8, dst, stride, above, left, goffset
-  GET_GOT     goffsetq
-  mova                  m0, [leftq]
-  movu                  m7, [aboveq-1]
-  movu                  m1, [aboveq+15]
-
-  pshufb                m4, m1, [GLOBAL(sh_b123456789abcdeff)]
-  pshufb                m6, m1, [GLOBAL(sh_b23456789abcdefff)]
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m1, m4, m6, m2          ; 3-tap avg above [high]
-
-  palignr               m3, m1, m7, 1
-  palignr               m5, m1, m7, 2
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m7, m3, m5, m1          ; 3-tap avg above [low]
-
-  pshufb                m7, [GLOBAL(sh_bfedcba9876543210)]
-  palignr               m5, m0, m7, 15
-  palignr               m3, m0, m7, 14
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m0, m5, m3, m4          ; 3-tap avg B3-Bg
-  pavgb                 m5, m0                            ; A1 - Ag
-  punpcklbw             m6, m4, m5                        ; A-B8 ... A-B1
-  punpckhbw             m4, m5                            ; A-B9 ... A-Bg
-  pshufb                m6, [GLOBAL(sh_bfedcba9876543210)]
-  pshufb                m4, [GLOBAL(sh_bfedcba9876543210)]
-
-  DEFINE_ARGS dst, stride, stride3, left, line
-  lea             stride3q, [strideq*3]
-
-  palignr               m5, m2, m1, 14
-  palignr               m7, m1, m6, 14
-  mova  [dstq            ], m7
-  mova  [dstq+16         ], m5
-  palignr               m5, m2, m1, 12
-  palignr               m7, m1, m6, 12
-  mova  [dstq+strideq    ], m7
-  mova  [dstq+strideq+16 ], m5
-  palignr                m5, m2, m1, 10
-  palignr                m7, m1, m6, 10
-  mova  [dstq+strideq*2   ], m7
-  mova  [dstq+strideq*2+16], m5
-  palignr                m5, m2, m1, 8
-  palignr                m7, m1, m6, 8
-  mova  [dstq+stride3q    ], m7
-  mova  [dstq+stride3q+16 ], m5
-  lea                  dstq, [dstq+strideq*4]
-  palignr                m5, m2, m1, 6
-  palignr                m7, m1, m6, 6
-  mova  [dstq             ], m7
-  mova  [dstq+16          ], m5
-  palignr                m5, m2, m1, 4
-  palignr                m7, m1, m6, 4
-  mova  [dstq+strideq     ], m7
-  mova  [dstq+strideq+16  ], m5
-  palignr                m5, m2, m1, 2
-  palignr                m7, m1, m6, 2
-  mova  [dstq+strideq*2   ], m7
-  mova  [dstq+strideq*2+16], m5
-  mova  [dstq+stride3q    ], m6
-  mova  [dstq+stride3q+16 ], m1
-  lea                  dstq, [dstq+strideq*4]
-
-  palignr                m5, m1, m6, 14
-  palignr                m3, m6, m4, 14
-  mova  [dstq             ], m3
-  mova  [dstq+16          ], m5
-  palignr                m5, m1, m6, 12
-  palignr                m3, m6, m4, 12
-  mova  [dstq+strideq     ], m3
-  mova  [dstq+strideq+16  ], m5
-  palignr                m5, m1, m6, 10
-  palignr                m3, m6, m4, 10
-  mova  [dstq+strideq*2   ], m3
-  mova  [dstq+strideq*2+16], m5
-  palignr                m5, m1, m6, 8
-  palignr                m3, m6, m4, 8
-  mova  [dstq+stride3q    ], m3
-  mova  [dstq+stride3q+16 ], m5
-  lea                  dstq, [dstq+strideq*4]
-  palignr                m5, m1, m6, 6
-  palignr                m3, m6, m4, 6
-  mova  [dstq             ], m3
-  mova  [dstq+16          ], m5
-  palignr                m5, m1, m6, 4
-  palignr                m3, m6, m4, 4
-  mova  [dstq+strideq     ], m3
-  mova  [dstq+strideq+16  ], m5
-  palignr                m5, m1, m6, 2
-  palignr                m3, m6, m4, 2
-  mova  [dstq+strideq*2   ], m3
-  mova  [dstq+strideq*2+16], m5
-  mova  [dstq+stride3q    ], m4
-  mova  [dstq+stride3q+16 ], m6
-  lea               dstq, [dstq+strideq*4]
-
-  mova                   m7, [leftq]
-  mova                   m3, [leftq+16]
-  palignr                m5, m3, m7, 15
-  palignr                m0, m3, m7, 14
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m3, m5, m0, m2          ; 3-tap avg Bh -
-  pavgb                  m5, m3                            ; Ah -
-  punpcklbw              m3, m2, m5                        ; A-B8 ... A-B1
-  punpckhbw              m2, m5                            ; A-B9 ... A-Bg
-  pshufb                 m3, [GLOBAL(sh_bfedcba9876543210)]
-  pshufb                 m2, [GLOBAL(sh_bfedcba9876543210)]
-
-  palignr                m7, m6, m4, 14
-  palignr                m0, m4, m3, 14
-  mova  [dstq             ], m0
-  mova  [dstq+16          ], m7
-  palignr                m7, m6, m4, 12
-  palignr                m0, m4, m3, 12
-  mova  [dstq+strideq     ], m0
-  mova  [dstq+strideq+16  ], m7
-  palignr                m7, m6, m4, 10
-  palignr                m0, m4, m3, 10
-  mova  [dstq+strideq*2   ], m0
-  mova  [dstq+strideq*2+16], m7
-  palignr                m7, m6, m4, 8
-  palignr                m0, m4, m3, 8
-  mova  [dstq+stride3q    ], m0
-  mova  [dstq+stride3q+16 ], m7
-  lea                  dstq, [dstq+strideq*4]
-  palignr                m7, m6, m4, 6
-  palignr                m0, m4, m3, 6
-  mova  [dstq             ], m0
-  mova  [dstq+16          ], m7
-  palignr                m7, m6, m4, 4
-  palignr                m0, m4, m3, 4
-  mova  [dstq+strideq     ], m0
-  mova  [dstq+strideq+16  ], m7
-  palignr                m7, m6, m4, 2
-  palignr                m0, m4, m3, 2
-  mova  [dstq+strideq*2   ], m0
-  mova  [dstq+strideq*2+16], m7
-  mova  [dstq+stride3q    ], m3
-  mova  [dstq+stride3q+16 ], m4
-  lea                  dstq, [dstq+strideq*4]
-
-  palignr                m7, m4, m3, 14
-  palignr                m0, m3, m2, 14
-  mova  [dstq             ], m0
-  mova  [dstq+16          ], m7
-  palignr                m7, m4, m3, 12
-  palignr                m0, m3, m2, 12
-  mova  [dstq+strideq     ], m0
-  mova  [dstq+strideq+16  ], m7
-  palignr                m7, m4, m3, 10
-  palignr                m0, m3, m2, 10
-  mova  [dstq+strideq*2   ], m0
-  mova  [dstq+strideq*2+16], m7
-  palignr                m7, m4, m3, 8
-  palignr                m0, m3, m2, 8
-  mova  [dstq+stride3q    ], m0
-  mova  [dstq+stride3q+16 ], m7
-  lea                  dstq, [dstq+strideq*4]
-  palignr                m7, m4, m3, 6
-  palignr                m0, m3, m2, 6
-  mova  [dstq             ], m0
-  mova  [dstq+16          ], m7
-  palignr                m7, m4, m3, 4
-  palignr                m0, m3, m2, 4
-  mova  [dstq+strideq     ], m0
-  mova  [dstq+strideq+16  ], m7
-  palignr                m7, m4, m3, 2
-  palignr                m0, m3, m2, 2
-  mova  [dstq+strideq*2   ], m0
-  mova  [dstq+strideq*2+16], m7
-  mova  [dstq+stride3q    ], m2
-  mova  [dstq+stride3q+16 ], m3
-
-  RESTORE_GOT
-  RET
-
-INIT_MMX ssse3
-cglobal d207_predictor_4x4, 4, 5, 4, dst, stride, unused, left, goffset
-  GET_GOT     goffsetq
-  movd                m0, [leftq]                ; abcd [byte]
-  pshufb              m1, m0, [GLOBAL(sh_b1233)] ; bcdd [byte]
-  pshufb              m3, m0, [GLOBAL(sh_b2333)] ; cddd
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m0, m1, m3, m2
-  pavgb               m1, m0             ; ab, bc, cd, d [byte]
-
-  punpcklbw           m1, m2             ; ab, a2bc, bc, b2cd, cd, c3d, d, d
-  movd    [dstq        ], m1
-  psrlq               m1, 16             ; bc, b2cd, cd, c3d, d, d
-  movd    [dstq+strideq], m1
-  lea               dstq, [dstq+strideq*2]
-  psrlq               m1, 16             ; cd, c3d, d, d
-  movd    [dstq        ], m1
-  pshufw              m1, m1, q1111      ; d, d, d, d
-  movd    [dstq+strideq], m1
-  RESTORE_GOT
-  RET
-
-INIT_XMM ssse3
-cglobal d207_predictor_8x8, 4, 5, 4, dst, stride, stride3, left, goffset
-  GET_GOT     goffsetq
-  movq                m3, [leftq]            ; abcdefgh [byte]
-  lea           stride3q, [strideq*3]
-
-  pshufb              m1, m3, [GLOBAL(sh_b2345677777777777)]
-  pshufb              m0, m3, [GLOBAL(sh_b0123456777777777)]
-  pshufb              m2, m3, [GLOBAL(sh_b1234567777777777)]
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m0, m2, m1, m3
-  pavgb               m0, m2
-  punpcklbw           m0, m3        ; interleaved output
-
-  movq  [dstq          ], m0
-  psrldq              m0, 2
-  movq  [dstq+strideq  ], m0
-  psrldq              m0, 2
-  movq  [dstq+strideq*2], m0
-  psrldq              m0, 2
-  movq  [dstq+stride3q ], m0
-  lea               dstq, [dstq+strideq*4]
-  pshufhw             m0, m0, q0000 ; de, d2ef, ef, e2fg, fg, f2gh, gh, g3h, 8xh
-  psrldq              m0, 2
-  movq  [dstq          ], m0
-  psrldq              m0, 2
-  movq  [dstq+strideq  ], m0
-  psrldq              m0, 2
-  movq  [dstq+strideq*2], m0
-  psrldq              m0, 2
-  movq  [dstq+stride3q ], m0
-  RESTORE_GOT
-  RET
-
-INIT_XMM ssse3
-cglobal d207_predictor_16x16, 4, 5, 5, dst, stride, stride3, left, goffset
-  GET_GOT     goffsetq
-  lea           stride3q, [strideq*3]
-  mova                m0, [leftq]            ; abcdefghijklmnop [byte]
-  pshufb              m1, m0, [GLOBAL(sh_b123456789abcdeff)] ; bcdefghijklmnopp
-  pshufb              m2, m0, [GLOBAL(sh_b23456789abcdefff)]
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m0, m1, m2, m3
-  pavgb               m1, m0                 ; ab, bc, cd .. no, op, pp [byte]
-
-  punpckhbw           m4, m1, m3    ; interleaved input
-  punpcklbw           m1, m3        ; interleaved output
-  mova  [dstq          ], m1
-  palignr             m3, m4, m1, 2
-  mova  [dstq+strideq  ], m3
-  palignr             m3, m4, m1, 4
-  mova  [dstq+strideq*2], m3
-  palignr             m3, m4, m1, 6
-  mova  [dstq+stride3q ], m3
-  lea               dstq, [dstq+strideq*4]
-  palignr             m3, m4, m1, 8
-  mova  [dstq          ], m3
-  palignr             m3, m4, m1, 10
-  mova  [dstq+strideq  ], m3
-  palignr             m3, m4, m1, 12
-  mova  [dstq+strideq*2], m3
-  palignr             m3, m4, m1, 14
-  mova  [dstq+stride3q ], m3
-  DEFINE_ARGS dst, stride, stride3, line
-  mov              lined, 2
-  mova                m0, [GLOBAL(sh_b23456789abcdefff)]
-.loop:
-  lea               dstq, [dstq+strideq*4]
-  mova  [dstq          ], m4
-  pshufb              m4, m0
-  mova  [dstq+strideq  ], m4
-  pshufb              m4, m0
-  mova  [dstq+strideq*2], m4
-  pshufb              m4, m0
-  mova  [dstq+stride3q ], m4
-  pshufb              m4, m0
-  dec              lined
-  jnz .loop
-  RESTORE_GOT
-  REP_RET
-
-INIT_XMM ssse3
-cglobal d207_predictor_32x32, 4, 5, 8, dst, stride, stride3, left, goffset
-  GET_GOT     goffsetq
-  lea           stride3q, [strideq*3]
-  mova                m1, [leftq]              ;  0-15 [byte]
-  mova                m2, [leftq+16]           ; 16-31 [byte]
-  pshufb              m0, m2, [GLOBAL(sh_b23456789abcdefff)]
-  pshufb              m4, m2, [GLOBAL(sh_b123456789abcdeff)]
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m2, m4, m0, m3
-  palignr             m6, m2, m1, 1
-  palignr             m5, m2, m1, 2
-  pavgb               m2, m4         ; high 16px even lines
-
-  X_PLUS_2Y_PLUS_Z_PLUS_2_RSH_2 m1, m6, m5, m0
-  pavgb                   m1, m6         ; low 16px even lines
-
-  punpckhbw               m6, m1, m0               ; interleaved output 2
-  punpcklbw               m1, m0                   ; interleaved output 1
-
-  punpckhbw               m7, m2, m3               ; interleaved output 4
-  punpcklbw               m2, m3                   ; interleaved output 3
-
-  ; output 1st 8 lines (and half of 2nd 8 lines)
-  DEFINE_ARGS dst, stride, stride3, dst8
-  lea                  dst8q, [dstq+strideq*8]
-  mova  [dstq              ], m1
-  mova  [dstq           +16], m6
-  mova  [dst8q             ], m6
-  palignr             m0, m6, m1, 2
-  palignr             m4, m2, m6, 2
-  mova  [dstq +strideq     ], m0
-  mova  [dstq +strideq  +16], m4
-  mova  [dst8q+strideq     ], m4
-  palignr             m0, m6, m1, 4
-  palignr             m4, m2, m6, 4
-  mova  [dstq +strideq*2   ], m0
-  mova  [dstq +strideq*2+16], m4
-  mova  [dst8q+strideq*2   ], m4
-  palignr             m0, m6, m1, 6
-  palignr             m4, m2, m6, 6
-  mova  [dstq +stride3q    ], m0
-  mova  [dstq +stride3q +16], m4
-  mova  [dst8q+stride3q    ], m4
-  lea               dstq, [dstq +strideq*4]
-  lea              dst8q, [dst8q+strideq*4]
-  palignr             m0, m6, m1, 8
-  palignr             m4, m2, m6, 8
-  mova  [dstq              ], m0
-  mova  [dstq           +16], m4
-  mova  [dst8q             ], m4
-  palignr             m0, m6, m1, 10
-  palignr             m4, m2, m6, 10
-  mova  [dstq +strideq     ], m0
-  mova  [dstq +strideq  +16], m4
-  mova  [dst8q+strideq     ], m4
-  palignr             m0, m6, m1, 12
-  palignr             m4, m2, m6, 12
-  mova  [dstq +strideq*2   ], m0
-  mova  [dstq +strideq*2+16], m4
-  mova  [dst8q+strideq*2   ], m4
-  palignr             m0, m6, m1, 14
-  palignr             m4, m2, m6, 14
-  mova  [dstq +stride3q    ], m0
-  mova  [dstq +stride3q +16], m4
-  mova  [dst8q+stride3q    ], m4
-  lea               dstq, [dstq+strideq*4]
-  lea              dst8q, [dst8q+strideq*4]
-
-  ; output 2nd half of 2nd 8 lines and half of 3rd 8 lines
-  mova  [dstq           +16], m2
-  mova  [dst8q             ], m2
-  palignr             m4, m7, m2, 2
-  mova  [dstq +strideq  +16], m4
-  mova  [dst8q+strideq     ], m4
-  palignr             m4, m7, m2, 4
-  mova  [dstq +strideq*2+16], m4
-  mova  [dst8q+strideq*2   ], m4
-  palignr             m4, m7, m2, 6
-  mova  [dstq +stride3q +16], m4
-  mova  [dst8q+stride3q    ], m4
-  lea               dstq, [dstq+strideq*4]
-  lea              dst8q, [dst8q+strideq*4]
-  palignr             m4, m7, m2, 8
-  mova  [dstq           +16], m4
-  mova  [dst8q             ], m4
-  palignr             m4, m7, m2, 10
-  mova  [dstq +strideq  +16], m4
-  mova  [dst8q+strideq     ], m4
-  palignr             m4, m7, m2, 12
-  mova  [dstq +strideq*2+16], m4
-  mova  [dst8q+strideq*2   ], m4
-  palignr             m4, m7, m2, 14
-  mova  [dstq +stride3q +16], m4
-  mova  [dst8q+stride3q    ], m4
-  lea               dstq, [dstq+strideq*4]
-  lea              dst8q, [dst8q+strideq*4]
-
-  ; output 2nd half of 3rd 8 lines and half of 4th 8 lines
-  mova                m0, [GLOBAL(sh_b23456789abcdefff)]
-  mova  [dstq           +16], m7
-  mova  [dst8q             ], m7
-  pshufb              m7, m0
-  mova  [dstq +strideq  +16], m7
-  mova  [dst8q+strideq     ], m7
-  pshufb              m7, m0
-  mova  [dstq +strideq*2+16], m7
-  mova  [dst8q+strideq*2   ], m7
-  pshufb              m7, m0
-  mova  [dstq +stride3q +16], m7
-  mova  [dst8q+stride3q    ], m7
-  pshufb              m7, m0
-  lea               dstq, [dstq+strideq*4]
-  lea              dst8q, [dst8q+strideq*4]
-  mova  [dstq           +16], m7
-  mova  [dst8q             ], m7
-  pshufb              m7, m0
-  mova  [dstq +strideq  +16], m7
-  mova  [dst8q+strideq     ], m7
-  pshufb              m7, m0
-  mova  [dstq +strideq*2+16], m7
-  mova  [dst8q+strideq*2   ], m7
-  pshufb              m7, m0
-  mova  [dstq +stride3q +16], m7
-  mova  [dst8q+stride3q    ], m7
-  pshufb              m7, m0
-  lea               dstq, [dstq+strideq*4]
-
-  ; output last half of 4th 8 lines
-  mova  [dstq           +16], m7
-  mova  [dstq +strideq  +16], m7
-  mova  [dstq +strideq*2+16], m7
-  mova  [dstq +stride3q +16], m7
-  lea               dstq, [dstq+strideq*4]
-  mova  [dstq           +16], m7
-  mova  [dstq +strideq  +16], m7
-  mova  [dstq +strideq*2+16], m7
-  mova  [dstq +stride3q +16], m7
-
-  ; done!
-  RESTORE_GOT
-  RET