Welcome to mirror list, hosted at ThFree Co, Russian Federation.

util_simd.h « util « cycles « intern - git.blender.org/blender.git - Unnamed repository; edit this file 'description' to name the repository.
summaryrefslogtreecommitdiff
blob: c8f794adfb39d031e705e3f188d17c43d087e19e (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
/*
 * Copyright 2011-2013 Blender Foundation
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 * http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License
 */

#ifndef __UTIL_SIMD_H__
#define __UTIL_SIMD_H__

CCL_NAMESPACE_BEGIN

#ifdef __KERNEL_SSE2__

/* SSE shuffle utility functions */

#ifdef __KERNEL_SSSE3__

/* faster version for SSSE3 */
typedef __m128i shuffle_swap_t;

ccl_device_inline const shuffle_swap_t shuffle_swap_identity(void)
{
	return _mm_set_epi8(15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0);
}

ccl_device_inline const shuffle_swap_t shuffle_swap_swap(void)
{
	return _mm_set_epi8(7, 6, 5, 4, 3, 2, 1, 0, 15, 14, 13, 12, 11, 10, 9, 8);
}

ccl_device_inline const __m128 shuffle_swap(const __m128& a, const shuffle_swap_t& shuf)
{
	return _mm_castsi128_ps(_mm_shuffle_epi8(_mm_castps_si128(a), shuf));
}

#else

/* somewhat slower version for SSE2 */
typedef int shuffle_swap_t;

ccl_device_inline const shuffle_swap_t shuffle_swap_identity(void)
{
	return 0;
}

ccl_device_inline const shuffle_swap_t shuffle_swap_swap(void)
{
	return 1;
}

ccl_device_inline const __m128 shuffle_swap(const __m128& a, shuffle_swap_t shuf)
{
	/* shuffle value must be a constant, so we need to branch */
	if(shuf)
		return _mm_shuffle_ps(a, a, _MM_SHUFFLE(1, 0, 3, 2));
	else
		return _mm_shuffle_ps(a, a, _MM_SHUFFLE(3, 2, 1, 0));
}

#endif

template<size_t i0, size_t i1, size_t i2, size_t i3> ccl_device_inline const __m128 shuffle(const __m128& a, const __m128& b)
{
	return _mm_shuffle_ps(a, b, _MM_SHUFFLE(i3, i2, i1, i0));
}

template<size_t i0, size_t i1, size_t i2, size_t i3> ccl_device_inline const __m128 shuffle(const __m128& a)
{
	return _mm_castsi128_ps(_mm_shuffle_epi32(_mm_castps_si128(a), _MM_SHUFFLE(i3, i2, i1, i0)));
}

template<size_t i0, size_t i1, size_t i2, size_t i3> ccl_device_inline const __m128i shuffle(const __m128i& a)
{
	return _mm_shuffle_epi32(a, _MM_SHUFFLE(i3, i2, i1, i0));
}

/* Blend 2 vectors based on mask: (a[i] & mask[i]) | (b[i] & ~mask[i]) */
#ifdef __KERNEL_SSE41__
ccl_device_inline const __m128 blend(const __m128& mask, const __m128& a, const __m128& b)
{
	return _mm_blendv_ps(b, a, mask);
}
#else
ccl_device_inline const __m128 blend(const __m128& mask, const __m128& a, const __m128& b)
{
	return _mm_or_ps(_mm_and_ps(mask, a), _mm_andnot_ps(mask, b));
}
#endif

/* calculate a*b+c (replacement for fused multiply-add on SSE CPUs) */
ccl_device_inline const __m128 fma(const __m128& a, const __m128& b, const __m128& c)
{
	return _mm_add_ps(_mm_mul_ps(a, b), c);
}

template<size_t N> ccl_device_inline const __m128 broadcast(const __m128& a)
{
	return _mm_castsi128_ps(_mm_shuffle_epi32(_mm_castps_si128(a), _MM_SHUFFLE(N, N, N, N)));
}

template<size_t N> ccl_device_inline const __m128i broadcast(const __m128i& a)
{
	return _mm_shuffle_epi32(a, _MM_SHUFFLE(N, N, N, N));
}

ccl_device_inline const __m128 uint32_to_float(const __m128i &in)
{
	__m128i a = _mm_srli_epi32(in, 16);
	__m128i b = _mm_and_si128(in, _mm_set1_epi32(0x0000ffff));
	__m128i c = _mm_or_si128(a, _mm_set1_epi32(0x53000000));
	__m128 d = _mm_cvtepi32_ps(b);
	__m128 e = _mm_sub_ps(_mm_castsi128_ps(c), _mm_castsi128_ps(_mm_set1_epi32(0x53000000)));
	return _mm_add_ps(e, d);
}

#endif /* __KERNEL_SSE2__ */

CCL_NAMESPACE_END

#endif /* __UTIL_SIMD_H__ */