Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(387)

Side by Side Diff: source/libvpx/third_party/libyuv/source/compare_posix.cc

Issue 1302353004: libvpx: Pull from upstream (Closed) Base URL: https://chromium.googlesource.com/chromium/deps/libvpx.git@master
Patch Set: Created 5 years, 3 months ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View unified diff | Download patch
OLDNEW
(Empty)
1 /*
2 * Copyright 2012 The LibYuv Project Authors. All rights reserved.
3 *
4 * Use of this source code is governed by a BSD-style license
5 * that can be found in the LICENSE file in the root of the source
6 * tree. An additional intellectual property rights grant can be found
7 * in the file PATENTS. All contributing project authors may
8 * be found in the AUTHORS file in the root of the source tree.
9 */
10
11 #include "libyuv/basic_types.h"
12 #include "libyuv/row.h"
13
14 #ifdef __cplusplus
15 namespace libyuv {
16 extern "C" {
17 #endif
18
19 #if !defined(LIBYUV_DISABLE_X86) && (defined(__x86_64__) || defined(__i386__))
20
21 uint32 SumSquareError_SSE2(const uint8* src_a, const uint8* src_b, int count) {
22 uint32 sse;
23 asm volatile ( // NOLINT
24 "pxor %%xmm0,%%xmm0 \n"
25 "pxor %%xmm5,%%xmm5 \n"
26 LABELALIGN
27 "1: \n"
28 "movdqu " MEMACCESS(0) ",%%xmm1 \n"
29 "lea " MEMLEA(0x10, 0) ",%0 \n"
30 "movdqu " MEMACCESS(1) ",%%xmm2 \n"
31 "lea " MEMLEA(0x10, 1) ",%1 \n"
32 "movdqa %%xmm1,%%xmm3 \n"
33 "psubusb %%xmm2,%%xmm1 \n"
34 "psubusb %%xmm3,%%xmm2 \n"
35 "por %%xmm2,%%xmm1 \n"
36 "movdqa %%xmm1,%%xmm2 \n"
37 "punpcklbw %%xmm5,%%xmm1 \n"
38 "punpckhbw %%xmm5,%%xmm2 \n"
39 "pmaddwd %%xmm1,%%xmm1 \n"
40 "pmaddwd %%xmm2,%%xmm2 \n"
41 "paddd %%xmm1,%%xmm0 \n"
42 "paddd %%xmm2,%%xmm0 \n"
43 "sub $0x10,%2 \n"
44 "jg 1b \n"
45
46 "pshufd $0xee,%%xmm0,%%xmm1 \n"
47 "paddd %%xmm1,%%xmm0 \n"
48 "pshufd $0x1,%%xmm0,%%xmm1 \n"
49 "paddd %%xmm1,%%xmm0 \n"
50 "movd %%xmm0,%3 \n"
51
52 : "+r"(src_a), // %0
53 "+r"(src_b), // %1
54 "+r"(count), // %2
55 "=g"(sse) // %3
56 :: "memory", "cc", "xmm0", "xmm1", "xmm2", "xmm3", "xmm5"
57 ); // NOLINT
58 return sse;
59 }
60
61 #endif // defined(__x86_64__) || defined(__i386__)
62
63 #if !defined(LIBYUV_DISABLE_X86) && \
64 (defined(__x86_64__) || (defined(__i386__) && !defined(__pic__)))
65 #define HAS_HASHDJB2_SSE41
66 static uvec32 kHash16x33 = { 0x92d9e201, 0, 0, 0 }; // 33 ^ 16
67 static uvec32 kHashMul0 = {
68 0x0c3525e1, // 33 ^ 15
69 0xa3476dc1, // 33 ^ 14
70 0x3b4039a1, // 33 ^ 13
71 0x4f5f0981, // 33 ^ 12
72 };
73 static uvec32 kHashMul1 = {
74 0x30f35d61, // 33 ^ 11
75 0x855cb541, // 33 ^ 10
76 0x040a9121, // 33 ^ 9
77 0x747c7101, // 33 ^ 8
78 };
79 static uvec32 kHashMul2 = {
80 0xec41d4e1, // 33 ^ 7
81 0x4cfa3cc1, // 33 ^ 6
82 0x025528a1, // 33 ^ 5
83 0x00121881, // 33 ^ 4
84 };
85 static uvec32 kHashMul3 = {
86 0x00008c61, // 33 ^ 3
87 0x00000441, // 33 ^ 2
88 0x00000021, // 33 ^ 1
89 0x00000001, // 33 ^ 0
90 };
91
92 uint32 HashDjb2_SSE41(const uint8* src, int count, uint32 seed) {
93 uint32 hash;
94 asm volatile ( // NOLINT
95 "movd %2,%%xmm0 \n"
96 "pxor %%xmm7,%%xmm7 \n"
97 "movdqa %4,%%xmm6 \n"
98 LABELALIGN
99 "1: \n"
100 "movdqu " MEMACCESS(0) ",%%xmm1 \n"
101 "lea " MEMLEA(0x10, 0) ",%0 \n"
102 "pmulld %%xmm6,%%xmm0 \n"
103 "movdqa %5,%%xmm5 \n"
104 "movdqa %%xmm1,%%xmm2 \n"
105 "punpcklbw %%xmm7,%%xmm2 \n"
106 "movdqa %%xmm2,%%xmm3 \n"
107 "punpcklwd %%xmm7,%%xmm3 \n"
108 "pmulld %%xmm5,%%xmm3 \n"
109 "movdqa %6,%%xmm5 \n"
110 "movdqa %%xmm2,%%xmm4 \n"
111 "punpckhwd %%xmm7,%%xmm4 \n"
112 "pmulld %%xmm5,%%xmm4 \n"
113 "movdqa %7,%%xmm5 \n"
114 "punpckhbw %%xmm7,%%xmm1 \n"
115 "movdqa %%xmm1,%%xmm2 \n"
116 "punpcklwd %%xmm7,%%xmm2 \n"
117 "pmulld %%xmm5,%%xmm2 \n"
118 "movdqa %8,%%xmm5 \n"
119 "punpckhwd %%xmm7,%%xmm1 \n"
120 "pmulld %%xmm5,%%xmm1 \n"
121 "paddd %%xmm4,%%xmm3 \n"
122 "paddd %%xmm2,%%xmm1 \n"
123 "paddd %%xmm3,%%xmm1 \n"
124 "pshufd $0xe,%%xmm1,%%xmm2 \n"
125 "paddd %%xmm2,%%xmm1 \n"
126 "pshufd $0x1,%%xmm1,%%xmm2 \n"
127 "paddd %%xmm2,%%xmm1 \n"
128 "paddd %%xmm1,%%xmm0 \n"
129 "sub $0x10,%1 \n"
130 "jg 1b \n"
131 "movd %%xmm0,%3 \n"
132 : "+r"(src), // %0
133 "+r"(count), // %1
134 "+rm"(seed), // %2
135 "=g"(hash) // %3
136 : "m"(kHash16x33), // %4
137 "m"(kHashMul0), // %5
138 "m"(kHashMul1), // %6
139 "m"(kHashMul2), // %7
140 "m"(kHashMul3) // %8
141 : "memory", "cc"
142 , "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7"
143 ); // NOLINT
144 return hash;
145 }
146 #endif // defined(__x86_64__) || (defined(__i386__) && !defined(__pic__)))
147
148 #ifdef __cplusplus
149 } // extern "C"
150 } // namespace libyuv
151 #endif
152
OLDNEW
« no previous file with comments | « source/libvpx/third_party/libyuv/source/compare_neon64.cc ('k') | source/libvpx/third_party/libyuv/source/compare_win.cc » ('j') | no next file with comments »

Powered by Google App Engine
This is Rietveld 408576698