Upgrade fwht4x4_mmx() to fwht4x4_sse2() for vp9 and vp10.

Function level timing test shows about 27% time saving on
a Xeon E5-2680 v2 desktop.

Rename vp9_dct_sse2.c to vp9_dct_intrin_sse2.c for vp9 and
rename dct_sse2.c to dct_intrin_sse2.c for vp10 to avoid
duplicate basenames.

Actually vp9_fwht4x4_mmx/sse2() and vp10_fwht4x4_mmx/sse2()
are identical. TODO: They should be unified later if there is
no intention to keep a duplicate.

Change-Id: I3e537b7bbd9ba417c606cd7c68c4dbbfa583f77d

3 files changed, 87 insertions, 104 deletions

diff --git a/vp9/encoder/x86/vp9_dct_sse2.c b/vp9/encoder/x86/vp9_dct_intrin_sse2.c
index fa37b6fed..fa37b6fed 100644
--- a/vp9/encoder/x86/vp9_dct_sse2.c
+++ b/vp9/encoder/x86/vp9_dct_intrin_sse2.c
diff --git a/vp9/encoder/x86/vp9_dct_mmx.asm b/vp9/encoder/x86/vp9_dct_mmx.asm
deleted file mode 100644
index 7a7a6b655..000000000
--- a/vp9/encoder/x86/vp9_dct_mmx.asm
+++ /dev/null
@@ -1,104 +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.
-;
-
-%define private_prefix vp9
-
-%include "third_party/x86inc/x86inc.asm"
-
-SECTION .text
-
-%macro TRANSFORM_COLS 0
-  paddw           m0,        m1
-  movq            m4,        m0
-  psubw           m3,        m2
-  psubw           m4,        m3
-  psraw           m4,        1
-  movq            m5,        m4
-  psubw           m5,        m1 ;b1
-  psubw           m4,        m2 ;c1
-  psubw           m0,        m4
-  paddw           m3,        m5
-                                ; m0 a0
-  SWAP            1,         4  ; m1 c1
-  SWAP            2,         3  ; m2 d1
-  SWAP            3,         5  ; m3 b1
-%endmacro
-
-%macro TRANSPOSE_4X4 0
-  movq            m4,        m0
-  movq            m5,        m2
-  punpcklwd       m4,        m1
-  punpckhwd       m0,        m1
-  punpcklwd       m5,        m3
-  punpckhwd       m2,        m3
-  movq            m1,        m4
-  movq            m3,        m0
-  punpckldq       m1,        m5
-  punpckhdq       m4,        m5
-  punpckldq       m3,        m2
-  punpckhdq       m0,        m2
-  SWAP            2, 3, 0, 1, 4
-%endmacro
-
-INIT_MMX mmx
-cglobal fwht4x4, 3, 4, 8, input, output, stride
-  lea             r3q,       [inputq + strideq*4]
-  movq            m0,        [inputq] ;a1
-  movq            m1,        [inputq + strideq*2] ;b1
-  movq            m2,        [r3q] ;c1
-  movq            m3,        [r3q + strideq*2] ;d1
-
-  TRANSFORM_COLS
-  TRANSPOSE_4X4
-  TRANSFORM_COLS
-  TRANSPOSE_4X4
-
-  psllw           m0,        2
-  psllw           m1,        2
-  psllw           m2,        2
-  psllw           m3,        2
-
-%if CONFIG_VP9_HIGHBITDEPTH
-  pxor            m4,             m4
-  pxor            m5,             m5
-  pcmpgtw         m4,             m0
-  pcmpgtw         m5,             m1
-  movq            m6,             m0
-  movq            m7,             m1
-  punpcklwd       m0,             m4
-  punpcklwd       m1,             m5
-  punpckhwd       m6,             m4
-  punpckhwd       m7,             m5
-  movq            [outputq],      m0
-  movq            [outputq + 8],  m6
-  movq            [outputq + 16], m1
-  movq            [outputq + 24], m7
-  pxor            m4,             m4
-  pxor            m5,             m5
-  pcmpgtw         m4,             m2
-  pcmpgtw         m5,             m3
-  movq            m6,             m2
-  movq            m7,             m3
-  punpcklwd       m2,             m4
-  punpcklwd       m3,             m5
-  punpckhwd       m6,             m4
-  punpckhwd       m7,             m5
-  movq            [outputq + 32], m2
-  movq            [outputq + 40], m6
-  movq            [outputq + 48], m3
-  movq            [outputq + 56], m7
-%else
-  movq            [outputq],      m0
-  movq            [outputq + 8],  m1
-  movq            [outputq + 16], m2
-  movq            [outputq + 24], m3
-%endif
-
-  RET
diff --git a/vp9/encoder/x86/vp9_dct_sse2.asm b/vp9/encoder/x86/vp9_dct_sse2.asm
new file mode 100644
index 000000000..d3b2a271b
--- /dev/null
+++ b/vp9/encoder/x86/vp9_dct_sse2.asm
@@ -0,0 +1,87 @@
+;
+;  Copyright (c) 2016 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.
+;
+
+%define private_prefix vp9
+
+%include "third_party/x86inc/x86inc.asm"
+
+SECTION .text
+
+%macro TRANSFORM_COLS 0
+  paddw           m0,        m1
+  movq            m4,        m0
+  psubw           m3,        m2
+  psubw           m4,        m3
+  psraw           m4,        1
+  movq            m5,        m4
+  psubw           m5,        m1 ;b1
+  psubw           m4,        m2 ;c1
+  psubw           m0,        m4
+  paddw           m3,        m5
+                                ; m0 a0
+  SWAP            1,         4  ; m1 c1
+  SWAP            2,         3  ; m2 d1
+  SWAP            3,         5  ; m3 b1
+%endmacro
+
+%macro TRANSPOSE_4X4 0
+                                ; 00 01 02 03
+                                ; 10 11 12 13
+                                ; 20 21 22 23
+                                ; 30 31 32 33
+  punpcklwd       m0,        m1 ; 00 10 01 11  02 12 03 13
+  punpcklwd       m2,        m3 ; 20 30 21 31  22 32 23 33
+  mova            m1,        m0
+  punpckldq       m0,        m2 ; 00 10 20 30  01 11 21 31
+  punpckhdq       m1,        m2 ; 02 12 22 32  03 13 23 33
+%endmacro
+
+INIT_XMM sse2
+cglobal fwht4x4, 3, 4, 8, input, output, stride
+; TODO(linfeng): The duplication with vp10 should be resolved.
+  lea             r3q,       [inputq + strideq*4]
+  movq            m0,        [inputq] ;a1
+  movq            m1,        [inputq + strideq*2] ;b1
+  movq            m2,        [r3q] ;c1
+  movq            m3,        [r3q + strideq*2] ;d1
+
+  TRANSFORM_COLS
+  TRANSPOSE_4X4
+  SWAP            1,         2
+  psrldq          m1,        m0, 8
+  psrldq          m3,        m2, 8
+  TRANSFORM_COLS
+  TRANSPOSE_4X4
+
+  psllw           m0,        2
+  psllw           m1,        2
+
+%if CONFIG_VP9_HIGHBITDEPTH
+  ; sign extension
+  mova            m2,             m0
+  mova            m3,             m1
+  punpcklwd       m0,             m0
+  punpcklwd       m1,             m1
+  punpckhwd       m2,             m2
+  punpckhwd       m3,             m3
+  psrad           m0,             16
+  psrad           m1,             16
+  psrad           m2,             16
+  psrad           m3,             16
+  mova            [outputq],      m0
+  mova            [outputq + 16], m2
+  mova            [outputq + 32], m1
+  mova            [outputq + 48], m3
+%else
+  mova            [outputq],      m0
+  mova            [outputq + 16], m1
+%endif
+
+  RET