/
benchmarkmerge.cu
181 lines (150 loc) · 6.17 KB
/
benchmarkmerge.cu
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
/******************************************************************************
* Copyright (c) 2013, NVIDIA CORPORATION. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * 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.
* * Neither the name of the NVIDIA CORPORATION 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 NVIDIA CORPORATION 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.
*
******************************************************************************/
/******************************************************************************
*
* Code and text by Sean Baxter, NVIDIA Research
* See http://nvlabs.github.io/moderngpu for repository and documentation.
*
******************************************************************************/
#include "kernels/merge.cuh"
#include <algorithm>
using namespace mgpu;
template<typename T>
void BenchmarkMergeKeys(int count, int numIt, CudaContext& context) {
#ifdef _DEBUG
numIt = 1;
#endif
int aCount = count / 2;
int bCount = count - aCount;
MGPU_MEM(T) a = context.SortRandom<T>(aCount, 0, (T)count);
MGPU_MEM(T) b = context.SortRandom<T>(bCount, 0, (T)count);
MGPU_MEM(T) c = context.Malloc<T>(count);
std::vector<T> aHost, bHost;
a->ToHost(aHost);
b->ToHost(bHost);
std::vector<T> cHost(count);
// Benchmark MGPU
context.Start();
for(int it = 0; it < numIt; ++it)
MergeKeys(a->get(), aCount, b->get(), bCount, c->get(), context);
double mgpuElapsed = context.Split();
// Benchmark STL
std::merge(aHost.begin(), aHost.end(), bHost.begin(), bHost.end(),
cHost.begin());
double cpuElapsed = context.Split();
// Compare MGPU to STL.
std::vector<T> cHost2;
c->ToHost(cHost2);
for(int i = 0; i < count; ++i)
if(cHost[i] != cHost2[i]) {
printf("MERGE ERROR AT COUNT = %d ITEM = %d\n", count, i);
exit(0);
}
double bytes = 2 * sizeof(T) * count;
double mgpuThroughput = count * numIt / mgpuElapsed;
double mgpuBandwidth = bytes * numIt / mgpuElapsed;
double cpuThroughput = count / cpuElapsed;
double cpuBandwidth = bytes / cpuElapsed;
printf("%s: %9.3lf M/s %7.3lf GB/s %9.3lf M/s %7.3lf GB/s\n",
FormatInteger(count).c_str(),
mgpuThroughput / 1.0e6, mgpuBandwidth / 1.0e9,
cpuThroughput / 1.0e6, cpuBandwidth / 1.0e9);
}
template<typename KeyType, typename ValType>
void BenchmarkMergePairs(int count, int numIt, CudaContext& context) {
#ifdef _DEBUG
numIt = 1;
#endif
int aCount = count / 2;
int bCount = count - aCount;
MGPU_MEM(KeyType) aKeys = context.SortRandom<KeyType>(aCount, 0, count);
MGPU_MEM(KeyType) bKeys = context.SortRandom<KeyType>(bCount, 0, count);
MGPU_MEM(KeyType) cKeys = context.Malloc<KeyType>(count);
MGPU_MEM(ValType) aVals = context.FillAscending<ValType>(aCount, 0, 1);
MGPU_MEM(ValType) bVals = context.FillAscending<ValType>(bCount, aCount, 1);
MGPU_MEM(ValType) cVals = context.Malloc<ValType>(count);
std::vector<KeyType> aHost, bHost;
aKeys->ToHost(aHost);
bKeys->ToHost(bHost);
// Benchmark MGPU.
context.Start();
for(int it = 0; it < numIt; ++it)
MergePairs(aKeys->get(), aVals->get(), aCount, bKeys->get(),
bVals->get(), bCount, cKeys->get(), cVals->get(),
mgpu::less<KeyType>(), context);
double mgpuElapsed = context.Split();
double bytes = 2 * (sizeof(KeyType) + sizeof(ValType)) * count;
double mgpuThroughput = count * numIt / mgpuElapsed;
double mgpuBandwidth = bytes * numIt / mgpuElapsed;
printf("%s: %9.3lf M/s %7.3lf GB/s\n",
FormatInteger(count).c_str(),
mgpuThroughput / 1.0e6, mgpuBandwidth / 1.0e9);
// Verify
std::vector<KeyType> keysHost;
std::vector<ValType> valsHost;
cKeys->ToHost(keysHost);
cVals->ToHost(valsHost);
for(int i = 0; i < count; ++i) {
ValType val = valsHost[i];
KeyType x = (val < aCount) ? aHost[val] : bHost[val - aCount];
if(x != keysHost[i]) {
printf("MISMATCH AT ELEMENT %d\n", i);
exit(0);
}
}
}
const int Tests[][2] = {
{ 10000, 1000 },
{ 50000, 1000 },
{ 100000, 1000 },
{ 200000, 500 },
{ 500000, 200 },
{ 1000000, 200 },
{ 2000000, 200 },
{ 5000000, 200 },
{ 10000000, 100 },
{ 20000000, 100 }
};
const int NumTests = sizeof(Tests) / sizeof(*Tests);
int main(int argc, char** argv) {
ContextPtr context = CreateCudaDevice(argc, argv, true);
typedef int T1;
typedef int64 T2;
printf("Benchmarking merge-keys on type %s.\n", TypeIdName<T1>());
for(int test = 0; test < NumTests; ++test)
BenchmarkMergeKeys<T1>(Tests[test][0], Tests[test][1], *context);
printf("\nBenchmarking merge-pairs on type %s.\n", TypeIdName<T1>());
for(int test = 0; test < NumTests; ++test)
BenchmarkMergePairs<T1, T1>(Tests[test][0], Tests[test][1], *context);
printf("\nBenchmarking merge-keys on type %s.\n", TypeIdName<T2>());
for(int test = 0; test < NumTests; ++test)
BenchmarkMergeKeys<T2>(Tests[test][0], Tests[test][1], *context);
printf("\nBenchmarking merge-pairs on type %s.\n", TypeIdName<T2>());
for(int test = 0; test < NumTests; ++test)
BenchmarkMergePairs<T2, T2>(Tests[test][0], Tests[test][1], *context);
return 0;
}