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
|
//---------------------------------------------------------------------------//
// Copyright (c) 2013 Kyle Lutz <kyle.r.lutz@gmail.com>
//
// Distributed under the Boost Software License, Version 1.0
// See accompanying file LICENSE_1_0.txt or copy at
// http://www.boost.org/LICENSE_1_0.txt
//
// See http://boostorg.github.com/compute for more information.
//---------------------------------------------------------------------------//
#define BOOST_TEST_MODULE TestDevice
#include <boost/test/unit_test.hpp>
#include <iostream>
#include <boost/compute/device.hpp>
#include <boost/compute/system.hpp>
#include <boost/compute/detail/nvidia_compute_capability.hpp>
#include "opencl_version_check.hpp"
BOOST_AUTO_TEST_CASE(null_device)
{
boost::compute::device null;
BOOST_CHECK(null.id() == cl_device_id());
BOOST_CHECK(null.get() == cl_device_id());
}
BOOST_AUTO_TEST_CASE(default_device_doctest)
{
//! [default_gpu]
boost::compute::device gpu = boost::compute::system::default_device();
//! [default_gpu]
BOOST_CHECK(gpu.id());
}
BOOST_AUTO_TEST_CASE(device_platform)
{
boost::compute::platform p = boost::compute::system::platforms().at(0);
BOOST_CHECK(p == p.devices().at(0).platform());
}
BOOST_AUTO_TEST_CASE(get_device_name)
{
boost::compute::device gpu = boost::compute::system::default_device();
if(gpu.id()){
BOOST_CHECK(!gpu.name().empty());
}
}
BOOST_AUTO_TEST_CASE(equality_operator)
{
boost::compute::device device1 = boost::compute::system::default_device();
BOOST_CHECK(device1 == device1);
boost::compute::device device2 = device1;
BOOST_CHECK(device1 == device2);
}
BOOST_AUTO_TEST_CASE(get_max_work_item_sizes)
{
boost::compute::device device = boost::compute::system::default_device();
std::vector<size_t> max_work_item_sizes =
device.get_info<std::vector<size_t> >(CL_DEVICE_MAX_WORK_ITEM_SIZES);
BOOST_CHECK_GE(max_work_item_sizes.size(), size_t(3));
BOOST_CHECK_GE(max_work_item_sizes[0], size_t(1));
BOOST_CHECK_GE(max_work_item_sizes[1], size_t(1));
BOOST_CHECK_GE(max_work_item_sizes[2], size_t(1));
}
#ifdef BOOST_COMPUTE_CL_VERSION_1_2
// returns true if the device supports the partitioning type
bool supports_partition_type(const boost::compute::device &device,
cl_device_partition_property type)
{
const std::vector<cl_device_partition_property> properties =
device.get_info<std::vector<cl_device_partition_property> >(
CL_DEVICE_PARTITION_PROPERTIES
);
return std::find(properties.begin(),
properties.end(),
type) != properties.end();
}
BOOST_AUTO_TEST_CASE(partition_device_equally)
{
// get default device and ensure it has at least two compute units
boost::compute::device device = boost::compute::system::default_device();
REQUIRES_OPENCL_VERSION(1,2);
if(device.compute_units() < 2){
std::cout << "skipping test: "
<< "device does not have enough compute units"
<< std::endl;
return;
}
// check that the device supports partitioning equally
if(!supports_partition_type(device, CL_DEVICE_PARTITION_EQUALLY)){
std::cout << "skipping test: "
<< "device does not support CL_DEVICE_PARTITION_EQUALLY"
<< std::endl;
return;
}
// ensure device is not a sub-device
BOOST_CHECK(device.is_subdevice() == false);
// partition default device into sub-devices with one compute unit each
std::vector<boost::compute::device>
sub_devices = device.partition_equally(1);
BOOST_CHECK_EQUAL(sub_devices.size(), size_t(device.compute_units()));
// verify each of the sub-devices
for(size_t i = 0; i < sub_devices.size(); i++){
const boost::compute::device &sub_device = sub_devices[i];
// ensure parent device id is correct
cl_device_id parent_id =
sub_device.get_info<cl_device_id>(CL_DEVICE_PARENT_DEVICE);
BOOST_CHECK(parent_id == device.id());
// ensure device is a sub-device
BOOST_CHECK(sub_device.is_subdevice() == true);
// check number of compute units
BOOST_CHECK_EQUAL(sub_device.compute_units(), size_t(1));
}
}
// used to sort devices by number of compute units
bool compare_compute_units(const boost::compute::device &a,
const boost::compute::device &b)
{
return a.compute_units() < b.compute_units();
}
BOOST_AUTO_TEST_CASE(partition_by_counts)
{
// get default device and ensure it has at least four compute units
boost::compute::device device = boost::compute::system::default_device();
REQUIRES_OPENCL_VERSION(1,2);
if(device.compute_units() < 4){
std::cout << "skipping test: "
<< "device does not have enough compute units"
<< std::endl;
return;
}
// check that the device supports partitioning by counts
if(!supports_partition_type(device, CL_DEVICE_PARTITION_BY_COUNTS)){
std::cout << "skipping test: "
<< "device does not support CL_DEVICE_PARTITION_BY_COUNTS"
<< std::endl;
return;
}
// ensure device is not a sub-device
BOOST_CHECK(device.is_subdevice() == false);
// create vector of sub-device compute unit counts
std::vector<size_t> counts;
counts.push_back(2);
counts.push_back(1);
counts.push_back(1);
// partition default device into sub-devices according to counts
std::vector<boost::compute::device>
sub_devices = device.partition_by_counts(counts);
BOOST_CHECK_EQUAL(sub_devices.size(), size_t(3));
// sort sub-devices by number of compute units (see issue #185)
std::sort(sub_devices.begin(), sub_devices.end(), compare_compute_units);
// verify each of the sub-devices
BOOST_CHECK_EQUAL(sub_devices[0].compute_units(), size_t(1));
BOOST_CHECK_EQUAL(sub_devices[1].compute_units(), size_t(1));
BOOST_CHECK_EQUAL(sub_devices[2].compute_units(), size_t(2));
}
BOOST_AUTO_TEST_CASE(partition_by_affinity_domain)
{
// get default device and ensure it has at least two compute units
boost::compute::device device = boost::compute::system::default_device();
REQUIRES_OPENCL_VERSION(1,2);
if(device.compute_units() < 2){
std::cout << "skipping test: "
<< "device does not have enough compute units"
<< std::endl;
return;
}
// check that the device supports splitting by affinity domains
if(!supports_partition_type(device, CL_DEVICE_AFFINITY_DOMAIN_NEXT_PARTITIONABLE)){
std::cout << "skipping test: "
<< "device does not support partitioning by affinity domain"
<< std::endl;
return;
}
// ensure device is not a sub-device
BOOST_CHECK(device.is_subdevice() == false);
std::vector<boost::compute::device> sub_devices =
device.partition_by_affinity_domain(
CL_DEVICE_AFFINITY_DOMAIN_NEXT_PARTITIONABLE);
BOOST_CHECK(sub_devices.size() > 0);
BOOST_CHECK(sub_devices[0].is_subdevice() == true);
}
#endif // BOOST_COMPUTE_CL_VERSION_1_2
BOOST_AUTO_TEST_CASE(nvidia_compute_capability)
{
boost::compute::device device = boost::compute::system::default_device();
int major, minor;
boost::compute::detail::get_nvidia_compute_capability(device, major, minor);
boost::compute::detail::check_nvidia_compute_capability(device, 3, 0);
}
BOOST_AUTO_TEST_CASE(get_info_specializations)
{
boost::compute::device device = boost::compute::system::default_device();
std::cout << device.get_info<CL_DEVICE_NAME>() << std::endl;
}
#ifdef BOOST_COMPUTE_CL_VERSION_2_1
BOOST_AUTO_TEST_CASE(get_host_timer)
{
boost::compute::device device = boost::compute::system::default_device();
REQUIRES_OPENCL_VERSION(2, 1);
BOOST_CHECK(device.get_host_timer() != 0);
#ifndef BOOST_COMPUTE_NO_HDR_CHRONO
typedef std::chrono::milliseconds stdms;
BOOST_CHECK(device.get_host_timer<stdms>().count() != 0);
#endif
#ifndef BOOST_COMPUTE_NO_BOOST_CHRONO
typedef boost::chrono::milliseconds bms;
BOOST_CHECK(device.get_host_timer<bms>().count() != 0);
#endif
}
BOOST_AUTO_TEST_CASE(get_device_and_host_timer)
{
boost::compute::device device = boost::compute::system::default_device();
REQUIRES_OPENCL_VERSION(2, 1);
typedef std::pair<boost::compute::ulong_, boost::compute::ulong_> dah_timer;
dah_timer timer;
BOOST_CHECK_NO_THROW(timer = device.get_device_and_host_timer());
BOOST_CHECK(timer.first != 0);
BOOST_CHECK(timer.second != 0);
#ifndef BOOST_COMPUTE_NO_HDR_CHRONO
typedef std::chrono::milliseconds stdms;
BOOST_CHECK(device.get_device_and_host_timer<stdms>().first.count() != 0);
BOOST_CHECK(device.get_device_and_host_timer<stdms>().second.count() != 0);
#endif
#ifndef BOOST_COMPUTE_NO_BOOST_CHRONO
typedef boost::chrono::milliseconds bms;
BOOST_CHECK(device.get_device_and_host_timer<bms>().first.count() != 0);
BOOST_CHECK(device.get_device_and_host_timer<bms>().second.count() != 0);
#endif
}
BOOST_AUTO_TEST_CASE(get_info_opencl21_queries)
{
boost::compute::device device = boost::compute::system::default_device();
REQUIRES_OPENCL_VERSION(2, 1);
BOOST_CHECK(!device.get_info<CL_DEVICE_IL_VERSION>().empty());
BOOST_CHECK(device.get_info<CL_DEVICE_MAX_NUM_SUB_GROUPS>() > 0);
BOOST_CHECK_NO_THROW(
device.get_info<CL_DEVICE_SUB_GROUP_INDEPENDENT_FORWARD_PROGRESS>()
);
}
#endif // BOOST_COMPUTE_CL_VERSION_2_1
|
