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
|
package com.mbien.opencl;
import com.mbien.opencl.CLMemory.Mem;
import com.mbien.opencl.CLMemory.GLObjectType;
import com.mbien.opencl.CLSampler.AddressingMode;
import com.mbien.opencl.CLSampler.FilteringMode;
import com.mbien.opencl.CLImageFormat.ChannelOrder;
import com.mbien.opencl.CLImageFormat.ChannelType;
import com.mbien.opencl.CLDevice.FPConfig;
import com.mbien.opencl.CLDevice.GlobalMemCacheType;
import com.mbien.opencl.CLDevice.LocalMemType;
import com.mbien.opencl.CLDevice.Type;
import java.io.IOException;
import java.nio.ByteBuffer;
import java.util.Arrays;
import java.util.EnumSet;
import java.util.Map;
import org.junit.BeforeClass;
import org.junit.Test;
import static org.junit.Assert.*;
import static java.lang.System.*;
import static com.mbien.opencl.TestUtils.*;
import static com.sun.gluegen.runtime.BufferFactory.*;
/**
* Test testing the high level bindings.
* @author Michael Bien
*/
public class HighLevelBindingTest {
@BeforeClass
public static void setUpClass() throws Exception {
out.println("OS: " + System.getProperty("os.name"));
out.println("ARCH: " + System.getProperty("os.arch"));
out.println("VM: " + System.getProperty("java.vm.name"));
out.println("lib path: " + System.getProperty("java.library.path"));
}
@Test
public void enumsTest() {
// enum tests
final EnumSet<FPConfig> singleFPConfig = FPConfig.valuesOf(CL.CL_FP_DENORM | CL.CL_FP_ROUND_TO_INF);
assertEquals(0, FPConfig.valuesOf(0).size());
assertTrue(singleFPConfig.contains(FPConfig.DENORM));
assertTrue(singleFPConfig.contains(FPConfig.ROUND_TO_INF));
// CLDevice enums
for (FPConfig e : FPConfig.values()) {
EnumSet<FPConfig> set = FPConfig.valuesOf(e.CONFIG);
assertTrue(set.contains(e));
}
for (GlobalMemCacheType e : GlobalMemCacheType.values()) {
assertEquals(e, GlobalMemCacheType.valueOf(e.TYPE));
}
for (LocalMemType e : LocalMemType.values()) {
assertEquals(e, LocalMemType.valueOf(e.TYPE));
}
for (Type e : Type.values()) {
assertEquals(e, Type.valueOf(e.TYPE));
}
// CLMemory enums
for (Mem e : Mem.values()) {
assertEquals(e, Mem.valueOf(e.CONFIG));
}
for (GLObjectType e : GLObjectType.values()) {
assertEquals(e, GLObjectType.valueOf(e.TYPE));
}
// CLSampler enums
for (AddressingMode e : AddressingMode.values()) {
assertEquals(e, AddressingMode.valueOf(e.MODE));
}
for (FilteringMode e : FilteringMode.values()) {
assertEquals(e, FilteringMode.valueOf(e.MODE));
}
// CLImage enums
for (ChannelOrder e : ChannelOrder.values()) {
assertEquals(e, ChannelOrder.valueOf(e.ORDER));
}
for (ChannelType e : ChannelType.values()) {
assertEquals(e, ChannelType.valueOf(e.TYPE));
}
}
@Test
public void contextlessTest() {
out.println(" - - - highLevelTest; contextless - - - ");
// platform/device info tests
CLPlatform[] clPlatforms = CLPlatform.listCLPlatforms();
for (CLPlatform platform : clPlatforms) {
out.println("platform info:");
out.println(" name: "+platform.getName());
out.println(" id: "+platform.ID);
out.println(" profile: "+platform.getProfile());
out.println(" version: "+platform.getVersion());
out.println(" vendor: "+platform.getVendor());
out.println(" max FLOPS device: "+platform.getMaxFlopsDevice());
out.println(" extensions: "+platform.getExtensions());
CLDevice[] clDevices = platform.listCLDevices();
for (CLDevice device : clDevices) {
out.println("device info:");
out.println(" name: "+device.getName());
out.println(" profile: "+device.getProfile());
out.println(" vendor: "+device.getVendor());
out.println(" vendor id: "+device.getVendorID());
out.println(" version: "+device.getVersion());
out.println(" driver version: "+device.getDriverVersion());
out.println(" type: "+device.getType());
out.println(" global mem: "+device.getGlobalMemSize()/(1024*1024)+" MB");
out.println(" max alloc mem: "+device.getMaxMemAllocSize()/(1024*1024)+" MB");
out.println(" max param size: "+device.getMaxParameterSize()+" byte");
out.println(" local mem: "+device.getLocalMemSize()/1024+" KB");
out.println(" local mem type: "+device.getLocalMemType());
out.println(" global mem cache size: "+device.getGlobalMemCacheSize());
out.println(" global mem cacheline size: "+device.getGlobalMemCachelineSize());
out.println(" global mem cache type: "+device.getGlobalMemCacheType());
out.println(" constant buffer size: "+device.getMaxConstantBufferSize());
out.println(" error correction support: "+device.isErrorCorrectionSupported());
out.println(" queue properties: "+device.getQueueProperties());
out.println(" clock: "+device.getMaxClockFrequency()+" MHz");
out.println(" timer res: "+device.getProfilingTimerResolution()+" ns");
out.println(" max work group size: "+device.getMaxWorkGroupSize());
out.println(" max compute units: "+device.getMaxComputeUnits());
out.println(" max work item dimensions: "+device.getMaxWorkItemDimensions());
out.println(" max work item sizes: "+Arrays.toString(device.getMaxWorkItemSizes()));
out.println(" compiler available: "+device.isCompilerAvailable());
out.println(" image support: "+device.isImageSupportAvailable());
out.println(" max read image args: "+device.getMaxReadImageArgs());
out.println(" max write image args: "+device.getMaxWriteImageArgs());
out.println(" max image2d dimensions: "+Arrays.asList(device.getMaxImage2dWidth(), device.getMaxImage2dHeight()));
out.println(" max image3d dimensions: "+Arrays.asList(device.getMaxImage2dWidth(), device.getMaxImage2dHeight(), device.getMaxImage3dDepth()));
out.println(" number of address bits: "+device.getAddressBits());
out.println(" half FP available: "+device.isHalfFPAvailable());
out.println(" double FP available: "+device.isDoubleFPAvailable());
out.println(" little endian: "+device.isLittleEndianAvailable());
out.println(" half FP config: "+device.getHalfFPConfig());
out.println(" single FP config: "+device.getSingleFPConfig());
out.println(" double FP config: "+device.getDoubleFPConfig());
out.println(" extensions: "+device.getExtensions());
}
}
}
@Test
public void createContextTest() {
out.println(" - - - highLevelTest; create context - - - ");
CLPlatform platform = CLPlatform.getDefault();
int deviceCount = platform.listCLDevices().length;
CLDevice firstDevice = platform.listCLDevices()[0];
CLContext c = CLContext.create();
assertNotNull(c);
assertEquals(deviceCount, c.getDevices().length);
c.release();
c = CLContext.create(platform);
assertNotNull(c);
assertEquals(deviceCount, c.getDevices().length);
c.release();
c = CLContext.create(firstDevice);
assertNotNull(c);
assertEquals(1, c.getDevices().length);
c.release();
c = CLContext.create(CLDevice.Type.ALL);
assertNotNull(c);
assertEquals(deviceCount, c.getDevices().length);
c.release();
c = CLContext.create(platform, firstDevice);
assertNotNull(c);
assertEquals(1, c.getDevices().length);
c.release();
c = CLContext.create(platform, CLDevice.Type.ALL);
assertNotNull(c);
assertEquals(deviceCount, c.getDevices().length);
c.release();
}
@Test
public void vectorAddGMTest() throws IOException {
out.println(" - - - highLevelTest; global memory kernel - - - ");
CLPlatform[] clPlatforms = CLPlatform.listCLPlatforms();
CLContext context = CLContext.create(clPlatforms[0]);
CLDevice[] contextDevices = context.getDevices();
out.println("context devices:");
for (CLDevice device : contextDevices) {
out.println(" "+device.toString());
}
out.println("max FLOPS device: " + context.getMaxFlopsDevice());
CLProgram program = context.createProgram(getClass().getResourceAsStream("testkernels.cl")).build();
CLDevice[] programDevices = program.getCLDevices();
assertEquals(contextDevices.length, programDevices.length);
out.println("build log:\n"+program.getBuildLog());
out.println("build status:\n"+program.getBuildStatus());
String source = program.getSource();
assertFalse(source.trim().isEmpty());
// out.println("source:\n"+source);
Map<CLDevice, byte[]> binaries = program.getBinaries();
assertFalse(binaries.isEmpty());
int elementCount = 11444777; // Length of float arrays to process (odd # for illustration)
int localWorkSize = 256; // set and log Global and Local work size dimensions
int globalWorkSize = roundUp(localWorkSize, elementCount); // rounded up to the nearest multiple of the LocalWorkSize
out.println("allocateing buffers of size: "+globalWorkSize);
ByteBuffer srcA = newDirectByteBuffer(globalWorkSize*SIZEOF_INT);
ByteBuffer srcB = newDirectByteBuffer(globalWorkSize*SIZEOF_INT);
ByteBuffer dest = newDirectByteBuffer(globalWorkSize*SIZEOF_INT);
fillBuffer(srcA, 23456);
fillBuffer(srcB, 46987);
CLBuffer<ByteBuffer> clBufferA = context.createBuffer(srcA, Mem.READ_ONLY);
CLBuffer<ByteBuffer> clBufferB = context.createBuffer(srcB, Mem.READ_ONLY);
CLBuffer<ByteBuffer> clBufferC = context.createBuffer(dest, Mem.WRITE_ONLY);
Map<String, CLKernel> kernels = program.createCLKernels();
for (CLKernel kernel : kernels.values()) {
out.println("kernel: "+kernel.toString());
}
assertNotNull(kernels.get("VectorAddGM"));
assertNotNull(kernels.get("Test"));
CLKernel vectorAddKernel = kernels.get("VectorAddGM");
vectorAddKernel.setArg(0, clBufferA)
.setArg(1, clBufferB)
.setArg(2, clBufferC)
.setArg(3, elementCount);
CLCommandQueue queue = programDevices[0].createCommandQueue();
// Asynchronous write of data to GPU device, blocking read later
queue.putWriteBuffer(clBufferA, false)
.putWriteBuffer(clBufferB, false)
.put1DRangeKernel(vectorAddKernel, 0, globalWorkSize, localWorkSize)
.putReadBuffer(clBufferC, true)
.finish().release();
out.println("a+b=c result snapshot: ");
for(int i = 0; i < 10; i++)
out.print(dest.getInt()+", ");
out.println("...; "+dest.remaining()/SIZEOF_INT + " more");
assertTrue(3 == context.getMemoryObjects().size());
clBufferA.release();
assertTrue(2 == context.getMemoryObjects().size());
assertTrue(2 == context.getMemoryObjects().size());
clBufferB.release();
assertTrue(1 == context.getMemoryObjects().size());
assertTrue(1 == context.getMemoryObjects().size());
clBufferC.release();
assertTrue(0 == context.getMemoryObjects().size());
assertTrue(1 == context.getPrograms().size());
program.release();
assertTrue(0 == context.getPrograms().size());
context.release();
}
}
|