forked from Rmalavally/rocm-examples
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmain.cpp
235 lines (207 loc) · 9.05 KB
/
main.cpp
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
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
// MIT License
//
// Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved.
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files (the "Software"), to deal
// in the Software without restriction, including without limitation the rights
// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
// copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included in all
// copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
// SOFTWARE.
#include "example_utils.hpp"
#include "hipblas_utils.hpp"
#include "hipsolver_utils.hpp"
#include <hipblas/hipblas.h>
#include <hipsolver/hipsolver.h>
#include <hip/hip_runtime.h>
#include <cstdlib>
#include <iostream>
#include <iterator>
#include <numeric>
#include <vector>
int main()
{
// Initialize dimensions, leading dimensions and number of elements of input matrices A and B.
constexpr unsigned int n = 3;
constexpr unsigned int lda = n;
constexpr unsigned int ldb = n;
constexpr unsigned int size_A = lda * n;
constexpr unsigned int size_B = ldb * n;
// Initialize symmetric input matrix A.
// | 0.5 3.5 0.0 |
// A = | 3.5 0.5 0.0 |
// | 0.0 0.0 2.0 |
std::vector<double> A{0.5, 3.5, 0.0, 3.5, 0.5, 0.0, 0.0, 0.0, 2.0};
// Initialize symmetric input matrix B as A but make it diagonally dominant
// https://en.wikipedia.org/wiki/Diagonally_dominant_matrix so it's positive definite.
// | 4.0 3.5 0.0 |
// B = | 3.5 4.0 0.0 |
// | 0.0 0.0 2.0 |
std::vector<double> B(A);
for(unsigned int i = 0; i < n; ++i)
{
double sum = 0;
for(unsigned int j = 0; j < n; ++j)
{
sum += std::fabs(B[i + j * ldb]);
}
B[i + i * ldb] = sum;
}
// Allocate device memory for the inputs and outputs and copy input matrices from host to device.
double* d_A{};
double* d_B{};
double* d_W{};
double* d_X{}; /*Auxiliary device matrix for solution checking.*/
int* d_sygvd_info{};
HIP_CHECK(hipMalloc(&d_A, sizeof(double) * size_A));
HIP_CHECK(hipMalloc(&d_B, sizeof(double) * size_B));
HIP_CHECK(hipMalloc(&d_W, sizeof(double) * n));
HIP_CHECK(hipMalloc(&d_X, sizeof(double) * size_A));
HIP_CHECK(hipMalloc(&d_sygvd_info, sizeof(int)));
HIP_CHECK(hipMemcpy(d_A, A.data(), sizeof(double) * size_A, hipMemcpyHostToDevice));
HIP_CHECK(hipMemcpy(d_B, B.data(), sizeof(double) * size_B, hipMemcpyHostToDevice));
// Use the hipSOLVER API to create a handle.
hipsolverHandle_t hipsolver_handle;
HIPSOLVER_CHECK(hipsolverCreate(&hipsolver_handle));
// Working space variables.
int lwork{};
double* d_work{};
// Query and allocate working space.
HIPSOLVER_CHECK(hipsolverDsygvd_bufferSize(hipsolver_handle,
HIPSOLVER_EIG_TYPE_1,
HIPSOLVER_EIG_MODE_VECTOR,
HIPSOLVER_FILL_MODE_UPPER,
n,
d_A,
lda,
d_B,
ldb,
d_W,
&lwork));
HIP_CHECK(hipMalloc(&d_work, lwork));
// Compute the eigenvalues (written to d_W) and eigenvectors (written to d_A) of the pair (A, B).
HIPSOLVER_CHECK(hipsolverDsygvd(hipsolver_handle,
HIPSOLVER_EIG_TYPE_1,
HIPSOLVER_EIG_MODE_VECTOR,
HIPSOLVER_FILL_MODE_UPPER,
n,
d_A,
lda,
d_B,
ldb,
d_W,
d_work,
lwork,
d_sygvd_info));
// Check output info value.
int sygvd_info{};
HIP_CHECK(hipMemcpy(&sygvd_info, d_sygvd_info, sizeof(sygvd_info), hipMemcpyDeviceToHost));
unsigned int errors{};
if(sygvd_info < 0)
{
std::cout << -sygvd_info << "-th parameter is wrong.\n" << std::endl;
errors++;
}
else if(sygvd_info > 0)
{
std::cout << "Computing eigenvalues did not converge.\n" << std::endl;
errors++;
}
else
{
std::cout << "Eigenvalues successfully computed: ";
// Copy the resulting vector of eigenvalues to the host and print it to standard output.
std::vector<double> W(n);
HIP_CHECK(hipMemcpy(W.data(), d_W, sizeof(double) * n, hipMemcpyDeviceToHost));
if(!W.empty())
{
std::copy(W.begin(),
std::prev(W.end()),
std::ostream_iterator<double>(std::cout, ", "));
std::cout << W.back() << "." << std::endl;
}
// Copy the resulting matrix of eigenvectors to the host.
std::vector<double> X(size_A);
HIP_CHECK(hipMemcpy(X.data(), d_A, sizeof(double) * size_A, hipMemcpyDeviceToHost));
// Check the solution using the hipBLAS API.
// Create a handle and enable passing scalar parameters from a pointer to host memory.
hipblasHandle_t hipblas_handle;
HIPBLAS_CHECK(hipblasCreate(&hipblas_handle));
HIPBLAS_CHECK(hipblasSetPointerMode(hipblas_handle, HIPBLAS_POINTER_MODE_HOST));
// Validate the result by seeing if B * X * W - A * X is the zero matrix.
const double eps = 1.0e5 * std::numeric_limits<double>::epsilon();
const double h_one = 1;
const double h_minus_one = -1;
const double h_zero{};
// Firstly, make A = A * X.
HIP_CHECK(hipMemcpy(d_X, X.data(), sizeof(double) * size_A, hipMemcpyHostToDevice));
HIP_CHECK(hipMemcpy(d_A, A.data(), sizeof(double) * size_A, hipMemcpyHostToDevice));
HIPBLAS_CHECK(hipblasDgemm(hipblas_handle,
HIPBLAS_OP_N,
HIPBLAS_OP_N,
n,
n,
n,
&h_one,
d_A,
lda,
d_X,
lda,
&h_zero,
d_A,
lda));
// Secondly, make X = X * diag(W).
HIP_CHECK(hipMemcpy(d_X, X.data(), sizeof(double) * size_A, hipMemcpyHostToDevice));
HIPBLAS_CHECK(
hipblasDdgmm(hipblas_handle, HIPBLAS_SIDE_RIGHT, n, n, d_X, lda, d_W, 1, d_X, lda));
// Thirdly, make A = B * X - A.
HIP_CHECK(hipMemcpy(d_B, B.data(), sizeof(double) * size_B, hipMemcpyHostToDevice));
HIPBLAS_CHECK(hipblasDgemm(hipblas_handle,
HIPBLAS_OP_N,
HIPBLAS_OP_N,
n,
n,
n,
&h_one,
d_B,
ldb,
d_X,
lda,
&h_minus_one,
d_A,
lda))
// Copy the result back to the host.
HIP_CHECK(hipMemcpy(A.data(), d_A, sizeof(double) * size_A, hipMemcpyDeviceToHost));
// Free hipBLAS handle.
HIPBLAS_CHECK(hipblasDestroy(hipblas_handle));
// Lastly, check if A is 0.
for(unsigned int j = 0; j < n; ++j)
{
for(unsigned int i = 0; i < n; ++i)
{
errors += std::fabs(A[i + j * lda]) > eps;
}
}
}
// Free resources.
HIP_CHECK(hipFree(d_A));
HIP_CHECK(hipFree(d_B));
HIP_CHECK(hipFree(d_W));
HIP_CHECK(hipFree(d_X));
HIP_CHECK(hipFree(d_work));
HIP_CHECK(hipFree(d_sygvd_info));
HIPSOLVER_CHECK(hipsolverDestroy(hipsolver_handle));
// Print validation result.
return report_validation_result(errors);
}