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
|
/*
* Copyright (c) 2017, Miroslav Stoyanov
*
* This file is part of
* Toolkit for Adaptive Stochastic Modeling And Non-Intrusive ApproximatioN: TASMANIAN
*
* Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
*
* 1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
*
* 2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions
* and the following disclaimer in the documentation and/or other materials provided with the distribution.
*
* 3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse
* or promote products derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES,
* INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY,
* OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA,
* OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
* OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
* UT-BATTELLE, LLC AND THE UNITED STATES GOVERNMENT MAKE NO REPRESENTATIONS AND DISCLAIM ALL WARRANTIES, BOTH EXPRESSED AND IMPLIED.
* THERE ARE NO EXPRESS OR IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, OR THAT THE USE OF THE SOFTWARE WILL NOT INFRINGE ANY PATENT,
* COPYRIGHT, TRADEMARK, OR OTHER PROPRIETARY RIGHTS, OR THAT THE SOFTWARE WILL ACCOMPLISH THE INTENDED RESULTS OR THAT THE SOFTWARE OR ITS USE WILL NOT RESULT IN INJURY OR DAMAGE.
* THE USER ASSUMES RESPONSIBILITY FOR ALL LIABILITIES, PENALTIES, FINES, CLAIMS, CAUSES OF ACTION, AND COSTS AND EXPENSES, CAUSED BY, RESULTING FROM OR ARISING OUT OF,
* IN WHOLE OR IN PART THE USE, STORAGE OR DISPOSAL OF THE SOFTWARE.
*/
#ifndef __TASMANIAN_SPARSE_GRID_ACCELERATED_DATA_STRUCTURES_CPP
#define __TASMANIAN_SPARSE_GRID_ACCELERATED_DATA_STRUCTURES_CPP
#include "tsgAcceleratedDataStructures.hpp"
namespace TasGrid{
std::map<std::string, TypeAcceleration> AccelerationMeta::getStringToAccelerationMap(){
return {
{"none", accel_none},
{"cpu-blas", accel_cpu_blas},
{"gpu-default", accel_gpu_default},
{"gpu-cublas", accel_gpu_cublas},
{"gpu-cuda", accel_gpu_cuda},
{"gpu-rocblas", accel_gpu_rocblas},
{"gpu-hip", accel_gpu_hip},
{"gpu-magma", accel_gpu_magma}};
}
TypeAcceleration AccelerationMeta::getIOAccelerationString(const char * name){
try{
return getStringToAccelerationMap().at(name);
}catch(std::out_of_range &){
return accel_none;
}
}
const char* AccelerationMeta::getIOAccelerationString(TypeAcceleration accel){
switch (accel){
case accel_cpu_blas: return "cpu-blas";
case accel_gpu_default: return "gpu-default";
case accel_gpu_cublas: return "gpu-cublas";
case accel_gpu_cuda: return "gpu-cuda";
case accel_gpu_magma: return "gpu-magma";
default: return "none";
}
}
int AccelerationMeta::getIOAccelerationInt(TypeAcceleration accel){
switch (accel){
case accel_cpu_blas: return 1;
case accel_gpu_default: return 3;
case accel_gpu_cublas: return 4;
case accel_gpu_cuda: return 5;
case accel_gpu_magma: return 6;
default: return 0;
}
}
TypeAcceleration AccelerationMeta::getIOIntAcceleration(int accel){
switch (accel){
case 1: return accel_cpu_blas;
case 3: return accel_gpu_default;
case 4: return accel_gpu_cublas;
case 5: return accel_gpu_cuda;
case 6: return accel_gpu_magma;
default: return accel_none;
}
}
bool AccelerationMeta::isAccTypeGPU(TypeAcceleration accel){
switch (accel){
case accel_gpu_default:
case accel_gpu_cublas:
case accel_gpu_cuda:
case accel_gpu_magma: return true;
default:
return false;
}
}
TypeAcceleration AccelerationMeta::getAvailableFallback(TypeAcceleration accel){
// sparse grids are evaluated in 2 stages:
// - s1: convert multi-index to matrix B
// - s2: multiply matrix B by stored matrix A
// Mode | Stage 1 device | Stage 2 device | Library for stage 2
// CUBLAS | CPU | GPU | Nvidia cuBlas (or cuSparse)
// CUDA | GPU | GPU | Nvidia cuBlas (or cuSparse)
// MAGMA | GPU* | GPU | UTK magma and magma_sparse
// BLAS | CPU | CPU | BLAS
// none | all done on CPU, still using OpenMP (if available)
// note that CUDA, HIP and DPCPP are interchangeable based on the selected backend at compiler time
#ifdef Tasmanian_ENABLE_DPCPP
// temporary workaround
// if (accel == accel_gpu_magma or accel == accel_gpu_cuda)
// return accel_gpu_cublas;
// return accel;
#endif
// accel_gpu_default should always point to the potentially "best" option (currently MAGMA)
if (accel == accel_gpu_default) accel = accel_gpu_magma;
#if !defined(Tasmanian_ENABLE_GPU) || !defined(Tasmanian_ENABLE_MAGMA) || !defined(Tasmanian_ENABLE_BLAS)
// if any of the 3 acceleration modes is missing, then add a switch statement to guard against setting that mode
switch(accel){
#ifndef Tasmanian_ENABLE_GPU
// if CUDA is missing: just use the CPU
case accel_gpu_cublas:
case accel_gpu_cuda:
#ifdef Tasmanian_ENABLE_BLAS
accel = accel_cpu_blas;
#else
accel = accel_none;
#endif
break;
#endif // Tasmanian_ENABLE_GPU
#ifndef Tasmanian_ENABLE_MAGMA
// MAGMA tries to use CUDA kernels with magma linear algebra, this CUDA is the next best thing
case accel_gpu_magma:
#ifdef Tasmanian_ENABLE_GPU
accel = accel_gpu_cuda;
#elif defined(Tasmanian_ENABLE_BLAS)
accel = accel_cpu_blas;
#else
accel = accel_none;
#endif
break;
#endif // Tasmanian_ENABLE_MAGMA
#ifndef Tasmanian_ENABLE_BLAS
// if BLAS is missing, do not attempt to use the GPU but go directly to "none" mode
case accel_cpu_blas:
accel = accel_none;
break;
#endif // Tasmanian_ENABLE_BLAS
default: // compiler complains if there is no explicit "default", even if empty
break;
}
#endif
return accel;
}
AccelerationDomainTransform::AccelerationDomainTransform(AccelerationContext const *acc, std::vector<double> const &transform_a, std::vector<double> const &transform_b) : acceleration(acc){
// The points are stored contiguously in a vector with stride equal to num_dimensions
// Using the contiguous memory in a contiguous fashion on the GPU implies that thread 0 works on dimension 0, thread 1 on dim 1 ...
// But the number of dimensions is often way less than the number of threads
// Therefore, we lump vectors together into large vectors of sufficient dimension
// The dimension is least 512, but less than max CUDA threads 1024
// The domain transforms are padded accordingly
num_dimensions = (int) transform_a.size();
padded_size = num_dimensions;
while(padded_size < 512) padded_size += num_dimensions;
std::vector<double> rate(padded_size);
std::vector<double> shift(padded_size);
int c = 0;
for(int i=0; i<padded_size; i++){
// instead of storing upper/lower limits (as in TasmanianSparseGrid) use rate and shift
double diff = transform_b[c] - transform_a[c];
rate[i] = 2.0 / diff;
shift[i] = (transform_b[c] + transform_a[c]) / diff;
c++;
c = (c % num_dimensions);
}
gpu_trans_a.load(acc, rate);
gpu_trans_b.load(acc, shift);
}
template<typename T>
void AccelerationDomainTransform::getCanonicalPoints(bool use01, const T *gpu_transformed_x, int num_x, GpuVector<T> &gpu_canonical_x){
gpu_canonical_x.resize(acceleration, ((size_t) num_dimensions) * ((size_t) num_x));
TasGpu::dtrans2can(acceleration, use01, num_dimensions, num_x, padded_size, gpu_trans_a.data(), gpu_trans_b.data(), gpu_transformed_x, gpu_canonical_x.data());
}
template void AccelerationDomainTransform::getCanonicalPoints<float>(bool, float const[], int, GpuVector<float>&);
template void AccelerationDomainTransform::getCanonicalPoints<double>(bool, double const[], int, GpuVector<double>&);
}
#endif
|
