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
|
/* Tests a case where buffers are >32bit large (mainly intended for Level Zero)
Copyright (c) 2024 Michal Babej / Intel Finland Oy
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 "pocl_opencl.h"
// Enable OpenCL C++ exceptions
#define CL_HPP_ENABLE_EXCEPTIONS
#define CL_HPP_MINIMUM_OPENCL_VERSION 120
#define CL_HPP_TARGET_OPENCL_VERSION 120
#include <CL/opencl.hpp>
#include <iostream>
#define INPUT_BUFFER_SIZE (5UL << 30)
#define OUTPUT_BUFFER_SIZE 128
#define WORKITEMS 1048576
// (INPUT_BUFFER_SIZE / WORKITEMS)
#define LOOPSIZE 5120
#define BUILD_OPTS "-DLOOPSIZE=5120"
const char *SOURCE = R"RAW(
__kernel void test_kernel2 (global const uchar *in,
global ulong *out)
{
size_t gid = get_global_id(0) * LOOPSIZE;
ulong sum = 0;
for (unsigned i = 0; i < LOOPSIZE; ++i) {
sum += in[gid + i];
}
atomic_fetch_add((volatile global atomic_ulong *)out, (ulong)sum);
};
)RAW";
int main(void) {
std::vector<cl::Platform> PlatformList;
bool Ok = false;
try {
// Pick platform
cl::Platform::get(&PlatformList);
cl::Platform UsedPlatform = PlatformList[0];
// Pick first platform
cl_context_properties CtxProps[] = {
CL_CONTEXT_PLATFORM, (cl_context_properties)(UsedPlatform()), 0};
cl::Context Context(CL_DEVICE_TYPE_ALL, CtxProps);
// Query the set of devices attched to the context
std::vector<cl::Device> Devices = Context.getInfo<CL_CONTEXT_DEVICES>();
cl::Device UsedDevice;
for (auto &Dev : Devices) {
cl_ulong MaxAlloc = Dev.getInfo<CL_DEVICE_MAX_MEM_ALLOC_SIZE>();
if (MaxAlloc > INPUT_BUFFER_SIZE) {
UsedDevice = Dev;
break;
}
}
if (UsedDevice() == nullptr) {
std::cout << "No device with sufficiently large MaxMemAlloc found.\n"
<< std::endl;
return 77;
} else {
std::cout << "Using device " << UsedDevice.getInfo<CL_DEVICE_NAME>()
<< " with MaxMemAllocSize: "
<< UsedDevice.getInfo<CL_DEVICE_MAX_MEM_ALLOC_SIZE>() << "\n";
}
unsigned char *Inputs = new unsigned char[INPUT_BUFFER_SIZE];
cl_ulong Sum = 0;
std::cout << "Generating random input...\n";
for (cl_ulong i = 0; i < (INPUT_BUFFER_SIZE / 4); i++) {
unsigned r = rand();
cl_ulong offset = i * 4;
memcpy(Inputs + offset, &r, sizeof(r));
Sum += Inputs[offset];
Sum += Inputs[offset + 1];
Sum += Inputs[offset + 2];
Sum += Inputs[offset + 3];
}
std::cout << "... done\n";
// Create and program from source
cl::Program::Sources Sources({SOURCE});
cl::Program Program(Context, Sources);
// Build program
Program.build(UsedDevice, BUILD_OPTS);
cl::Buffer InBuffer = cl::Buffer(Context,
CL_MEM_READ_ONLY | CL_MEM_COPY_HOST_PTR |
CL_MEM_ALLOC_HOST_PTR,
INPUT_BUFFER_SIZE, (void *)&Inputs[0]);
cl::Buffer OutBuffer =
cl::Buffer(Context, CL_MEM_WRITE_ONLY, OUTPUT_BUFFER_SIZE);
cl::Kernel Kernel(Program, "test_kernel2");
Kernel.setArg(0, InBuffer);
Kernel.setArg(1, OutBuffer);
cl::CommandQueue Queue(Context, Devices[0], 0);
// clear output buffer
cl_ulong Zeroes = 0;
Queue.enqueueFillBuffer(OutBuffer, Zeroes, 0, OUTPUT_BUFFER_SIZE);
Queue.enqueueNDRangeKernel(Kernel, cl::NullRange, cl::NDRange(WORKITEMS),
cl::NullRange);
cl_ulong *Output =
(cl_ulong *)Queue.enqueueMapBuffer(OutBuffer,
CL_TRUE, // block
CL_MAP_READ, 0, OUTPUT_BUFFER_SIZE);
// verify
Ok = (Output[0] == Sum);
Queue.enqueueUnmapMemObject(OutBuffer, (void *)Output);
Queue.finish();
} catch (cl::Error &err) {
std::cerr << "ERROR: " << err.what() << "(" << err.err() << ")"
<< std::endl;
return EXIT_FAILURE;
}
PlatformList[0].unloadCompiler();
if (Ok) {
std::cout << "Compare OK" << std::endl;
return EXIT_SUCCESS;
} else {
std::cout << "FAIL: result mismatch" << std::endl;
return EXIT_FAILURE;
}
}
|
