• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 //
2 // Copyright (c) 2017 The Khronos Group Inc.
3 //
4 // Licensed under the Apache License, Version 2.0 (the "License");
5 // you may not use this file except in compliance with the License.
6 // You may obtain a copy of the License at
7 //
8 //    http://www.apache.org/licenses/LICENSE-2.0
9 //
10 // Unless required by applicable law or agreed to in writing, software
11 // distributed under the License is distributed on an "AS IS" BASIS,
12 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 // See the License for the specific language governing permissions and
14 // limitations under the License.
15 //
16 #include "testBase.h"
17 #include "harness/typeWrappers.h"
18 #include "harness/testHarness.h"
19 #include "harness/conversions.h"
20 
21 const char *test_kernels[] = { "__kernel void kernelA(__global uint *dst)\n"
22                                "{\n"
23                                "\n"
24                                " dst[get_global_id(0)]*=3;\n"
25                                "\n"
26                                "}\n"
27                                "__kernel void kernelB(__global uint *dst)\n"
28                                "{\n"
29                                "\n"
30                                " dst[get_global_id(0)]++;\n"
31                                "\n"
32                                "}\n" };
33 
34 #define TEST_SIZE    512
35 #define MAX_DEVICES 32
36 #define MAX_QUEUES 1000
37 
test_device_set(size_t deviceCount,size_t queueCount,cl_device_id * devices,int num_elements)38 int test_device_set(size_t deviceCount, size_t queueCount, cl_device_id *devices, int num_elements)
39 {
40     int error;
41     clContextWrapper context;
42     clProgramWrapper program;
43     clKernelWrapper kernels[2];
44     clMemWrapper      stream;
45     clCommandQueueWrapper queues[MAX_QUEUES] = {};
46     size_t    threads[1], localThreads[1];
47     cl_uint data[TEST_SIZE];
48     cl_uint outputData[TEST_SIZE];
49     cl_uint expectedResults[TEST_SIZE];
50     cl_uint expectedResultsOneDevice[MAX_DEVICES][TEST_SIZE];
51     size_t i;
52 
53     RandomSeed seed( gRandomSeed );
54 
55     if (deviceCount > MAX_DEVICES) {
56        log_error("Number of devices in set (%ld) is greater than the number for which the test was written (%d).", deviceCount, MAX_DEVICES);
57     return -1;
58   }
59 
60   if (queueCount > MAX_QUEUES) {
61        log_error("Number of queues (%ld) is greater than the number for which the test was written (%d).", queueCount, MAX_QUEUES);
62     return -1;
63   }
64 
65   log_info("Testing with %ld queues on %ld devices, %ld kernel executions.\n", queueCount, deviceCount, queueCount*num_elements/TEST_SIZE);
66 
67   for (i=0; i<deviceCount; i++) {
68     char deviceName[4096] = "";
69     error = clGetDeviceInfo(devices[i], CL_DEVICE_NAME, sizeof(deviceName), deviceName, NULL);
70     test_error(error, "clGetDeviceInfo CL_DEVICE_NAME failed");
71     log_info("Device %ld is \"%s\".\n", i, deviceName);
72   }
73 
74     /* Create a context */
75     context = clCreateContext( NULL, (cl_uint)deviceCount, devices, notify_callback, NULL, &error );
76     test_error( error, "Unable to create testing context" );
77 
78     /* Create our kernels (they all have the same arguments so we don't need multiple ones for each device) */
79   if( create_single_kernel_helper( context, &program, &kernels[0], 1, test_kernels, "kernelA" ) != 0 )
80   {
81     return -1;
82   }
83 
84   kernels[1] = clCreateKernel(program, "kernelB", &error);
85   test_error(error, "clCreateKernel failed");
86 
87 
88     /* Now create I/O streams */
89   for( i = 0; i < TEST_SIZE; i++ )
90     data[i] = genrand_int32(seed);
91 
92   stream = clCreateBuffer(context, CL_MEM_COPY_HOST_PTR,
93                           sizeof(cl_uint) * TEST_SIZE, data, &error);
94   test_error(error, "Unable to create test array");
95 
96   // Update the expected results
97   for( i = 0; i < TEST_SIZE; i++ ) {
98     expectedResults[i] = data[i];
99     for (size_t j=0; j<deviceCount; j++)
100       expectedResultsOneDevice[j][i] = data[i];
101   }
102 
103 
104   // Set the arguments
105   error = clSetKernelArg( kernels[0], 0, sizeof( stream ), &stream);
106   test_error( error, "Unable to set kernel arguments" );
107   error = clSetKernelArg( kernels[1], 0, sizeof( stream ), &stream);
108   test_error( error, "Unable to set kernel arguments" );
109 
110     /* Run the test */
111     threads[0] = (size_t)TEST_SIZE;
112 
113     error = get_max_common_work_group_size( context, kernels[0], threads[0], &localThreads[ 0 ] );
114     test_error( error, "Unable to calc work group size" );
115 
116     /* Create work queues */
117     for( i = 0; i < queueCount; i++ )
118     {
119         queues[i] = clCreateCommandQueue( context, devices[ i % deviceCount ], 0, &error );
120     if (error != CL_SUCCESS || queues[i] == NULL) {
121       log_info("Could not create queue[%d].\n", (int)i);
122       queueCount = i;
123       break;
124     }
125     }
126   log_info("Testing with %d queues.\n", (int)queueCount);
127 
128     /* Enqueue executions */
129   for( int z = 0; z<num_elements/TEST_SIZE; z++) {
130     for( i = 0; i < queueCount; i++ )
131     {
132       // Randomly choose a kernel to execute.
133       int kernel_selection = (int)get_random_float(0, 2, seed);
134       error = clEnqueueNDRangeKernel( queues[ i ], kernels[ kernel_selection ], 1, NULL, threads, localThreads, 0, NULL, NULL );
135       test_error( error, "Kernel execution failed" );
136 
137       // Update the expected results
138       for( int j = 0; j < TEST_SIZE; j++ ) {
139         expectedResults[j] = (kernel_selection) ? expectedResults[j]+1 : expectedResults[j]*3;
140         expectedResultsOneDevice[i % deviceCount][j] = (kernel_selection) ? expectedResultsOneDevice[i % deviceCount][j]+1 : expectedResultsOneDevice[i % deviceCount][j]*3;
141       }
142 
143       // Force the queue to finish so the next one will be in sync
144       error = clFinish(queues[i]);
145       test_error( error, "clFinish failed");
146     }
147   }
148 
149   /* Read results */
150   int errors = 0;
151   for (int q = 0; q<(int)queueCount; q++) {
152     error = clEnqueueReadBuffer( queues[ 0 ], stream, CL_TRUE, 0, sizeof(cl_int)*TEST_SIZE, (char *)outputData, 0, NULL, NULL );
153     test_error( error, "Unable to get result data set" );
154 
155     int errorsThisTime = 0;
156     /* Verify all of the data now */
157     for( i = 0; i < TEST_SIZE; i++ )
158     {
159       if( expectedResults[ i ] != outputData[ i ] )
160       {
161         log_error( "ERROR: Sample data did not verify for queue %d on device %ld (sample %d, expected %d, got %d)\n",
162                   q, q % deviceCount, (int)i, expectedResults[ i ], outputData[ i ] );
163         for (size_t j=0; j<deviceCount; j++) {
164           if (expectedResultsOneDevice[j][i] == outputData[i])
165             log_info("Sample consistent with only device %ld having modified the data.\n", j);
166         }
167         errorsThisTime++;
168         break;
169       }
170     }
171     if (errorsThisTime)
172       errors++;
173   }
174 
175     /* All done now! */
176   if (errors) return -1;
177   return 0;
178 }
179 
test_two_devices(cl_device_id deviceID,cl_context context,cl_command_queue queue,int num_elements)180 int test_two_devices(cl_device_id deviceID, cl_context context, cl_command_queue queue, int num_elements)
181 {
182     cl_platform_id platform;
183     cl_device_id devices[2];
184     int err;
185     cl_uint numDevices;
186 
187     err = clGetPlatformIDs(1, &platform, NULL);
188     test_error( err, "Unable to get platform" );
189 
190     /* Get some devices */
191     err = clGetDeviceIDs(platform,  CL_DEVICE_TYPE_ALL, 2, devices, &numDevices );
192     test_error( err, "Unable to get 2 devices" );
193 
194     if( numDevices < 2 )
195     {
196         log_info( "WARNING: two device test unable to get two devices via CL_DEVICE_TYPE_ALL (got %d devices). Skipping test...\n", (int)numDevices );
197         return 0;
198     }
199   else if (numDevices > 2)
200   {
201     log_info("Note: got %d devices, using just the first two.\n", (int)numDevices);
202   }
203 
204     /* Run test */
205     return test_device_set( 2, 2, devices, num_elements );
206 }
207 
test_max_devices(cl_device_id deviceID,cl_context context,cl_command_queue queue,int num_elements)208 int test_max_devices(cl_device_id deviceID, cl_context context, cl_command_queue queue, int num_elements)
209 {
210     cl_platform_id platform;
211     cl_device_id devices[MAX_DEVICES];
212     cl_uint deviceCount;
213     int err;
214 
215     err = clGetPlatformIDs(1, &platform, NULL);
216     test_error( err, "Unable to get platform" );
217 
218     /* Get some devices */
219     err = clGetDeviceIDs(platform,  CL_DEVICE_TYPE_ALL, MAX_DEVICES, devices, &deviceCount );
220     test_error( err, "Unable to get multiple devices" );
221 
222   log_info("Testing with %d devices.", deviceCount);
223 
224     /* Run test */
225     return test_device_set( deviceCount, deviceCount, devices, num_elements );
226 }
227 
test_hundred_queues(cl_device_id device,cl_context contextIgnore,cl_command_queue queueIgnore,int num_elements)228 int test_hundred_queues(cl_device_id device, cl_context contextIgnore, cl_command_queue queueIgnore, int num_elements)
229 {
230   return test_device_set( 1, 100, &device, num_elements );
231 }
232 
233