|
1 /* |
|
2 * Copyright (c) 2005 |
|
3 * Eric Anholt. All rights reserved. |
|
4 * |
|
5 * Redistribution and use in source and binary forms, with or without |
|
6 * modification, are permitted provided that the following conditions |
|
7 * are met: |
|
8 * 1. Redistributions of source code must retain the above copyright |
|
9 * notice, this list of conditions and the following disclaimer. |
|
10 * 2. Redistributions in binary form must reproduce the above copyright |
|
11 * notice, this list of conditions and the following disclaimer in the |
|
12 * documentation and/or other materials provided with the distribution. |
|
13 * |
|
14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND |
|
15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
|
16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
|
17 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE |
|
18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
|
19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
|
20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
|
21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
|
22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
|
23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
|
24 * SUCH DAMAGE. |
|
25 */ |
|
26 //Portions Copyright (c) 2008-2009 Nokia Corporation and/or its subsidiary(-ies). All rights reserved. |
|
27 |
|
28 #ifdef HAVE_CONFIG_H |
|
29 #include "config.h" |
|
30 #endif |
|
31 #include <liboil/liboilclasses.h> |
|
32 #include <liboil/liboilfunction.h> |
|
33 |
|
34 #ifdef HAVE_GCC_ASM |
|
35 #include <mmintrin.h> |
|
36 #endif |
|
37 |
|
38 static void |
|
39 splat_u32_ns_mmx (uint32_t *dest, const uint32_t *param, int n) |
|
40 { |
|
41 #ifdef HAVE_GCC_ASM |
|
42 __m64 v; |
|
43 |
|
44 v = _mm_set1_pi32(*param); |
|
45 |
|
46 for (; n >= 2; n -= 2) { |
|
47 *(__m64 *)dest = v; |
|
48 dest += 2; |
|
49 } |
|
50 for (; n > 0; n--) { |
|
51 *dest++ = *param; |
|
52 } |
|
53 _mm_empty(); |
|
54 #endif |
|
55 } |
|
56 OIL_DEFINE_IMPL_FULL (splat_u32_ns_mmx, splat_u32_ns, OIL_IMPL_FLAG_MMX); |
|
57 |
|
58 static void |
|
59 splat_u32_ns_mmx_unroll4 (uint32_t *dest, const uint32_t *param, int n) |
|
60 { |
|
61 #ifdef HAVE_GCC_ASM |
|
62 __m64 v; |
|
63 |
|
64 v = _mm_set1_pi32(*param); |
|
65 |
|
66 for (; n >= 8; n -= 8) { |
|
67 ((__m64 *)dest)[0] = v; |
|
68 ((__m64 *)dest)[1] = v; |
|
69 ((__m64 *)dest)[2] = v; |
|
70 ((__m64 *)dest)[3] = v; |
|
71 dest += 8; |
|
72 } |
|
73 for (; n >= 2; n -= 2) { |
|
74 *(__m64 *)dest = v; |
|
75 dest += 2; |
|
76 } |
|
77 for (; n > 0; n--) { |
|
78 *dest++ = *param; |
|
79 } |
|
80 _mm_empty(); |
|
81 #endif |
|
82 } |
|
83 OIL_DEFINE_IMPL_FULL (splat_u32_ns_mmx_unroll4, splat_u32_ns, OIL_IMPL_FLAG_MMX); |
|
84 |
|
85 static void |
|
86 splat_u8_ns_mmx (uint8_t *dest, const uint8_t *param, int n) |
|
87 { |
|
88 #ifdef HAVE_GCC_ASM |
|
89 __m64 v; |
|
90 |
|
91 v = _mm_set1_pi8(*param); |
|
92 |
|
93 for (; n >= 8; n -= 8) { |
|
94 *(__m64 *)dest = v; |
|
95 dest += 8; |
|
96 } |
|
97 for (; n > 0; n--) { |
|
98 *dest++ = *param; |
|
99 } |
|
100 _mm_empty(); |
|
101 #endif |
|
102 } |
|
103 OIL_DEFINE_IMPL_FULL (splat_u8_ns_mmx, splat_u8_ns, OIL_IMPL_FLAG_MMX); |
|
104 |
|
105 static void |
|
106 splat_u8_ns_mmx_unroll4 (uint8_t *dest, const uint8_t *param, int n) |
|
107 { |
|
108 #ifdef HAVE_GCC_ASM |
|
109 __m64 v; |
|
110 |
|
111 v = _mm_set1_pi8(*param); |
|
112 |
|
113 for (; n >= 32; n -= 32) { |
|
114 ((__m64 *)dest)[0] = v; |
|
115 ((__m64 *)dest)[1] = v; |
|
116 ((__m64 *)dest)[2] = v; |
|
117 ((__m64 *)dest)[3] = v; |
|
118 dest += 32; |
|
119 } |
|
120 for (; n >= 8; n -= 8) { |
|
121 *(__m64 *)dest = v; |
|
122 dest += 8; |
|
123 } |
|
124 for (; n > 0; n--) { |
|
125 *dest++ = *param; |
|
126 } |
|
127 _mm_empty(); |
|
128 #endif |
|
129 } |
|
130 OIL_DEFINE_IMPL_FULL (splat_u8_ns_mmx_unroll4, splat_u8_ns, OIL_IMPL_FLAG_MMX); |
|
131 |
|
132 |
|
133 #ifdef __SYMBIAN32__ |
|
134 |
|
135 OilFunctionImpl* __oil_function_impl_splat_u32_ns_mmx() { |
|
136 return &_oil_function_impl_splat_u32_ns_mmx; |
|
137 } |
|
138 #endif |
|
139 |
|
140 #ifdef __SYMBIAN32__ |
|
141 |
|
142 OilFunctionImpl* __oil_function_impl_splat_u32_ns_mmx_unroll4() { |
|
143 return &_oil_function_impl_splat_u32_ns_mmx_unroll4; |
|
144 } |
|
145 #endif |
|
146 |
|
147 #ifdef __SYMBIAN32__ |
|
148 |
|
149 OilFunctionImpl* __oil_function_impl_splat_u8_ns_mmx() { |
|
150 return &_oil_function_impl_splat_u8_ns_mmx; |
|
151 } |
|
152 #endif |
|
153 |
|
154 #ifdef __SYMBIAN32__ |
|
155 |
|
156 OilFunctionImpl* __oil_function_impl_splat_u8_ns_mmx_unroll4() { |
|
157 return &_oil_function_impl_splat_u8_ns_mmx_unroll4; |
|
158 } |
|
159 #endif |
|
160 |