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
|
/*
* Copyright 2008-2009 NVIDIA Corporation
*
* 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.
*/
#pragma once
#include <cusp/detail/device/arch.h>
#include <cusp/detail/device/utils.h>
#include <cusp/detail/device/dereference.h>
#include <thrust/iterator/iterator_traits.h>
namespace cusp
{
namespace detail
{
namespace device
{
namespace cuda
{
template <int BLOCK_SIZE,
typename SizeType,
typename IndexIterator1,
typename IndexIterator2,
typename ValueIterator1,
typename ValueIterator2,
typename ValueIterator3,
typename ValueIterator4,
typename BinaryFunction1,
typename BinaryFunction2>
__launch_bounds__(BLOCK_SIZE,1)
__global__
void spmv_csr_scalar_kernel(SizeType num_rows,
IndexIterator1 row_offsets,
IndexIterator2 column_indices,
ValueIterator1 values,
ValueIterator2 x,
ValueIterator3 y,
ValueIterator4 z,
BinaryFunction1 combine,
BinaryFunction2 reduce)
{
typedef typename thrust::iterator_value<IndexIterator1>::type IndexType1;
typedef typename thrust::iterator_value<IndexIterator2>::type IndexType2;
typedef typename thrust::iterator_value<ValueIterator1>::type ValueType1;
typedef typename thrust::iterator_value<ValueIterator2>::type ValueType2;
typedef typename thrust::iterator_value<ValueIterator3>::type ValueType3;
typedef typename thrust::iterator_value<ValueIterator4>::type ValueType4;
const SizeType thread_id = BLOCK_SIZE * blockIdx.x + threadIdx.x;
const SizeType grid_size = BLOCK_SIZE * gridDim.x;
for(SizeType i = thread_id; i < num_rows; i += grid_size)
{
IndexIterator1 r0 = row_offsets; r0 += i; IndexType1 row_start = CUSP_DEREFERENCE(r0); // row_offsets[i]
IndexIterator1 r1 = row_offsets; r1 += i + 1; IndexType1 row_end = CUSP_DEREFERENCE(r1); // row_offsets[i + 1]
ValueIterator3 y0 = y; y0 += i; ValueType4 sum = CUSP_DEREFERENCE(y0); // sum = y[i]
for (IndexType2 jj = row_start; jj < row_end; jj++)
{
IndexIterator2 c0 = column_indices; c0 += jj; IndexType2 j = CUSP_DEREFERENCE(c0); // j = column_indices[jj]
ValueIterator1 v0 = values; v0 += jj; ValueType1 A_ij = CUSP_DEREFERENCE(v0); // A_ij = values[jj]
ValueIterator2 x0 = x; x0 += j; ValueType2 x_j = CUSP_DEREFERENCE(x0); // x_j = x[j]
sum = reduce(sum, combine(A_ij, x_j)); // sum += A_ij * x_j
}
ValueIterator4 z0 = z; z0 += i; CUSP_DEREFERENCE(z0) = sum; // z[i] = sum
}
}
template <typename SizeType,
typename IndexIterator1,
typename IndexIterator2,
typename ValueIterator1,
typename ValueIterator2,
typename ValueIterator3,
typename ValueIterator4,
typename BinaryFunction1,
typename BinaryFunction2>
void spmv_csr_scalar(SizeType num_rows,
IndexIterator1 row_offsets,
IndexIterator2 column_indices,
ValueIterator1 values,
ValueIterator2 x,
ValueIterator3 y,
ValueIterator4 z,
BinaryFunction1 combine,
BinaryFunction2 reduce)
{
const SizeType block_size = 256;
const SizeType max_blocks = cusp::detail::device::arch::max_active_blocks(spmv_csr_scalar_kernel<block_size, SizeType, IndexIterator1, IndexIterator2, ValueIterator1, ValueIterator2, ValueIterator3, ValueIterator4, BinaryFunction1, BinaryFunction2>, block_size, (size_t) 0);
const SizeType num_blocks = std::min(max_blocks, DIVIDE_INTO(num_rows, block_size));
spmv_csr_scalar_kernel<block_size><<<num_blocks, block_size>>>
(num_rows,
row_offsets, column_indices, values,
x, y, z,
combine, reduce);
}
} // end namespace cuda
} // end namespace device
} // end namespace detail
} // end namespace cusp
|
