-
Notifications
You must be signed in to change notification settings - Fork 39
/
TRAP_INT-Cuda.cpp
313 lines (211 loc) · 8.06 KB
/
TRAP_INT-Cuda.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
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
//~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~//
// Copyright (c) 2017-24, Lawrence Livermore National Security, LLC
// and RAJA Performance Suite project contributors.
// See the RAJAPerf/LICENSE file for details.
//
// SPDX-License-Identifier: (BSD-3-Clause)
//~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~//
#include "TRAP_INT.hpp"
#include "RAJA/RAJA.hpp"
#if defined(RAJA_ENABLE_CUDA)
#include "TRAP_INT-func.hpp"
#include "common/CudaDataUtils.hpp"
#include <iostream>
#include <utility>
#include <type_traits>
#include <limits>
namespace rajaperf
{
namespace basic
{
template < size_t block_size >
__launch_bounds__(block_size)
__global__ void trapint(Real_type x0, Real_type xp,
Real_type y, Real_type yp,
Real_type h,
Real_ptr sumx,
Index_type iend)
{
extern __shared__ Real_type psumx[ ];
Index_type i = blockIdx.x * block_size + threadIdx.x;
psumx[ threadIdx.x ] = 0.0;
for ( ; i < iend ; i += gridDim.x * block_size ) {
Real_type x = x0 + i*h;
Real_type val = trap_int_func(x, y, xp, yp);
psumx[ threadIdx.x ] += val;
}
__syncthreads();
for ( i = block_size / 2; i > 0; i /= 2 ) {
if ( threadIdx.x < i ) {
psumx[ threadIdx.x ] += psumx[ threadIdx.x + i ];
}
__syncthreads();
}
if ( threadIdx.x == 0 ) {
RAJA::atomicAdd<RAJA::cuda_atomic>( sumx, psumx[ 0 ] );
}
}
template < size_t block_size, typename MappingHelper >
void TRAP_INT::runCudaVariantBase(VariantID vid)
{
const Index_type run_reps = getRunReps();
const Index_type iend = getActualProblemSize();
auto res{getCudaResource()};
TRAP_INT_DATA_SETUP;
if ( vid == Base_CUDA ) {
RAJAPERF_CUDA_REDUCER_SETUP(Real_ptr, sumx, hsumx, 1, 1);
constexpr size_t shmem = sizeof(Real_type)*block_size;
const size_t max_grid_size = RAJAPERF_CUDA_GET_MAX_BLOCKS(
MappingHelper, (trapint<block_size>), block_size, shmem);
startTimer();
for (RepIndex_type irep = 0; irep < run_reps; ++irep) {
RAJAPERF_CUDA_REDUCER_INITIALIZE(&m_sumx_init, sumx, hsumx, 1, 1);
const size_t normal_grid_size = RAJA_DIVIDE_CEILING_INT(iend, block_size);
const size_t grid_size = std::min(normal_grid_size, max_grid_size);
RPlaunchCudaKernel( (trapint<block_size>),
grid_size, block_size,
shmem, res.get_stream(),
x0, xp,
y, yp,
h,
sumx,
iend);
RAJAPERF_CUDA_REDUCER_COPY_BACK(sumx, hsumx, 1, 1);
m_sumx += hsumx[0] * h;
}
stopTimer();
RAJAPERF_CUDA_REDUCER_TEARDOWN(sumx, hsumx);
} else {
getCout() << "\n TRAP_INT : Unknown Cuda variant id = " << vid << std::endl;
}
}
template < size_t block_size, typename AlgorithmHelper, typename MappingHelper >
void TRAP_INT::runCudaVariantRAJA(VariantID vid)
{
using reduction_policy = std::conditional_t<AlgorithmHelper::atomic,
RAJA::cuda_reduce_atomic,
RAJA::cuda_reduce>;
using exec_policy = std::conditional_t<MappingHelper::direct,
RAJA::cuda_exec<block_size, true /*async*/>,
RAJA::cuda_exec_occ_calc<block_size, true /*async*/>>;
const Index_type run_reps = getRunReps();
const Index_type ibegin = 0;
const Index_type iend = getActualProblemSize();
auto res{getCudaResource()};
TRAP_INT_DATA_SETUP;
if ( vid == RAJA_CUDA ) {
startTimer();
for (RepIndex_type irep = 0; irep < run_reps; ++irep) {
RAJA::ReduceSum<reduction_policy, Real_type> sumx(m_sumx_init);
RAJA::forall<exec_policy>( res,
RAJA::RangeSegment(ibegin, iend), [=] __device__ (Index_type i) {
TRAP_INT_BODY;
});
m_sumx += static_cast<Real_type>(sumx.get()) * h;
}
stopTimer();
} else {
getCout() << "\n TRAP_INT : Unknown Cuda variant id = " << vid << std::endl;
}
}
template < size_t block_size, typename MappingHelper >
void TRAP_INT::runCudaVariantRAJANewReduce(VariantID vid)
{
using exec_policy = std::conditional_t<MappingHelper::direct,
RAJA::cuda_exec<block_size, true /*async*/>,
RAJA::cuda_exec_occ_calc<block_size, true /*async*/>>;
const Index_type run_reps = getRunReps();
const Index_type ibegin = 0;
const Index_type iend = getActualProblemSize();
auto res{getCudaResource()};
TRAP_INT_DATA_SETUP;
if ( vid == RAJA_CUDA ) {
startTimer();
for (RepIndex_type irep = 0; irep < run_reps; ++irep) {
Real_type tsumx = m_sumx_init;
RAJA::forall<exec_policy>( res,
RAJA::RangeSegment(ibegin, iend),
RAJA::expt::Reduce<RAJA::operators::plus>(&tsumx),
[=] __device__ (Index_type i, Real_type& sumx) {
TRAP_INT_BODY;
}
);
m_sumx += static_cast<Real_type>(tsumx) * h;
}
stopTimer();
} else {
getCout() << "\n TRAP_INT : Unknown Cuda variant id = " << vid << std::endl;
}
}
void TRAP_INT::runCudaVariant(VariantID vid, size_t tune_idx)
{
size_t t = 0;
if ( vid == Base_CUDA || vid == RAJA_CUDA ) {
seq_for(gpu_block_sizes_type{}, [&](auto block_size) {
if (run_params.numValidGPUBlockSize() == 0u ||
run_params.validGPUBlockSize(block_size)) {
seq_for(gpu_mapping::reducer_helpers{}, [&](auto mapping_helper) {
if ( vid == Base_CUDA ) {
if (tune_idx == t) {
setBlockSize(block_size);
runCudaVariantBase<decltype(block_size){},
decltype(mapping_helper)>(vid);
}
t += 1;
} else if ( vid == RAJA_CUDA ) {
seq_for(gpu_algorithm::reducer_helpers{}, [&](auto algorithm_helper) {
if (tune_idx == t) {
setBlockSize(block_size);
runCudaVariantRAJA<decltype(block_size){},
decltype(algorithm_helper),
decltype(mapping_helper)>(vid);
}
t += 1;
});
if (tune_idx == t) {
setBlockSize(block_size);
runCudaVariantRAJANewReduce<decltype(block_size){},
decltype(mapping_helper)>(vid);
}
t += 1;
}
});
}
});
} else {
getCout() << "\n TRAP_INT : Unknown Cuda variant id = " << vid << std::endl;
}
}
void TRAP_INT::setCudaTuningDefinitions(VariantID vid)
{
if ( vid == Base_CUDA || vid == RAJA_CUDA ) {
seq_for(gpu_block_sizes_type{}, [&](auto block_size) {
if (run_params.numValidGPUBlockSize() == 0u ||
run_params.validGPUBlockSize(block_size)) {
seq_for(gpu_mapping::reducer_helpers{}, [&](auto mapping_helper) {
if ( vid == Base_CUDA ) {
auto algorithm_helper = gpu_algorithm::block_atomic_helper{};
addVariantTuningName(vid, decltype(algorithm_helper)::get_name()+"_"+
decltype(mapping_helper)::get_name()+"_"+
std::to_string(block_size));
RAJA_UNUSED_VAR(algorithm_helper); // to quiet compiler warning
} else if ( vid == RAJA_CUDA ) {
seq_for(gpu_algorithm::reducer_helpers{}, [&](auto algorithm_helper) {
addVariantTuningName(vid, decltype(algorithm_helper)::get_name()+"_"+
decltype(mapping_helper)::get_name()+"_"+
std::to_string(block_size));
});
auto algorithm_helper = gpu_algorithm::block_device_helper{};
addVariantTuningName(vid, decltype(algorithm_helper)::get_name()+"_"+
decltype(mapping_helper)::get_name()+"_"+
"new_"+std::to_string(block_size));
RAJA_UNUSED_VAR(algorithm_helper); // to quiet compiler warning
}
});
}
});
}
}
} // end namespace basic
} // end namespace rajaperf
#endif // RAJA_ENABLE_CUDA