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 "harness/compat.h"
17
18 #include <stdio.h>
19 #include <string.h>
20 #include <sys/types.h>
21 #include <sys/stat.h>
22
23 #include "procs.h"
24
25 static int
26 test_sign_double(cl_device_id device, cl_context context, cl_command_queue queue, int n_elems);
27
28
29 const char *sign_kernel_code =
30 "__kernel void test_sign(__global float *src, __global float *dst)\n"
31 "{\n"
32 " int tid = get_global_id(0);\n"
33 "\n"
34 " dst[tid] = sign(src[tid]);\n"
35 "}\n";
36
37 const char *sign2_kernel_code =
38 "__kernel void test_sign2(__global float2 *src, __global float2 *dst)\n"
39 "{\n"
40 " int tid = get_global_id(0);\n"
41 "\n"
42 " dst[tid] = sign(src[tid]);\n"
43 "}\n";
44
45 const char *sign4_kernel_code =
46 "__kernel void test_sign4(__global float4 *src, __global float4 *dst)\n"
47 "{\n"
48 " int tid = get_global_id(0);\n"
49 "\n"
50 " dst[tid] = sign(src[tid]);\n"
51 "}\n";
52
53 const char *sign8_kernel_code =
54 "__kernel void test_sign8(__global float8 *src, __global float8 *dst)\n"
55 "{\n"
56 " int tid = get_global_id(0);\n"
57 "\n"
58 " dst[tid] = sign(src[tid]);\n"
59 "}\n";
60
61 const char *sign16_kernel_code =
62 "__kernel void test_sign16(__global float16 *src, __global float16 *dst)\n"
63 "{\n"
64 " int tid = get_global_id(0);\n"
65 "\n"
66 " dst[tid] = sign(src[tid]);\n"
67 "}\n";
68
69 const char *sign3_kernel_code =
70 "__kernel void test_sign3(__global float *src, __global float *dst)\n"
71 "{\n"
72 " int tid = get_global_id(0);\n"
73 "\n"
74 " vstore3(sign(vload3(tid,src)), tid, dst);\n"
75 "}\n";
76
77
78
79 static int
verify_sign(float * inptr,float * outptr,int n)80 verify_sign(float *inptr, float *outptr, int n)
81 {
82 float r;
83 int i;
84
85 for (i=0; i<n; i++)
86 {
87 if (inptr[i] > 0.0f)
88 r = 1.0f;
89 else if (inptr[i] < 0.0f)
90 r = -1.0f;
91 else
92 r = 0.0f;
93 if (r != outptr[i])
94 return -1;
95 }
96
97 return 0;
98 }
99
100 static const char *fn_names[] = { "SIGN float", "SIGN float2", "SIGN float4", "SIGN float8", "SIGN float16", "SIGN float3" };
101
102 int
test_sign(cl_device_id device,cl_context context,cl_command_queue queue,int n_elems)103 test_sign(cl_device_id device, cl_context context, cl_command_queue queue, int n_elems)
104 {
105 cl_mem streams[2];
106 cl_float *input_ptr[1], *output_ptr, *p;
107 cl_program program[kTotalVecCount];
108 cl_kernel kernel[kTotalVecCount];
109 void *values[2];
110 size_t threads[1];
111 int num_elements;
112 int err;
113 int i;
114 MTdata d;
115
116 num_elements = n_elems * 16;
117
118 input_ptr[0] = (cl_float*)malloc(sizeof(cl_float) * num_elements);
119 output_ptr = (cl_float*)malloc(sizeof(cl_float) * num_elements);
120 streams[0] = clCreateBuffer(context, CL_MEM_READ_WRITE,
121 sizeof(cl_float) * num_elements, NULL, NULL);
122 if (!streams[0])
123 {
124 log_error("clCreateBuffer failed\n");
125 return -1;
126 }
127
128 streams[1] = clCreateBuffer(context, CL_MEM_READ_WRITE,
129 sizeof(cl_float) * num_elements, NULL, NULL);
130 if (!streams[1])
131 {
132 log_error("clCreateBuffer failed\n");
133 return -1;
134 }
135
136 d = init_genrand( gRandomSeed );
137 p = input_ptr[0];
138 for (i=0; i<num_elements; i++)
139 {
140 p[i] = get_random_float(-0x20000000, 0x20000000, d);
141 }
142 free_mtdata(d); d = NULL;
143
144
145 err = clEnqueueWriteBuffer( queue, streams[0], true, 0, sizeof(cl_float)*num_elements, (void *)input_ptr[0], 0, NULL, NULL );
146 if (err != CL_SUCCESS)
147 {
148 log_error("clWriteArray failed\n");
149 return -1;
150 }
151
152 err = create_single_kernel_helper( context, &program[0], &kernel[0], 1, &sign_kernel_code, "test_sign" );
153 if (err)
154 return -1;
155 err = create_single_kernel_helper( context, &program[1], &kernel[1], 1, &sign2_kernel_code, "test_sign2" );
156 if (err)
157 return -1;
158 err = create_single_kernel_helper( context, &program[2], &kernel[2], 1, &sign4_kernel_code, "test_sign4" );
159 if (err)
160 return -1;
161 err = create_single_kernel_helper( context, &program[3], &kernel[3], 1, &sign8_kernel_code, "test_sign8" );
162 if (err)
163 return -1;
164 err = create_single_kernel_helper( context, &program[4], &kernel[4], 1, &sign16_kernel_code, "test_sign16" );
165 if (err)
166 return -1;
167 err = create_single_kernel_helper( context, &program[5], &kernel[5], 1, &sign3_kernel_code, "test_sign3" );
168 if (err)
169 return -1;
170
171 values[0] = streams[0];
172 values[1] = streams[1];
173 for (i=0; i<kTotalVecCount; i++)
174 {
175 err = clSetKernelArg(kernel[i], 0, sizeof streams[0], &streams[0] );
176 err |= clSetKernelArg(kernel[i], 1, sizeof streams[1], &streams[1] );
177 if (err != CL_SUCCESS)
178 {
179 log_error("clSetKernelArgs failed\n");
180 return -1;
181 }
182 }
183
184 threads[0] = (size_t)n_elems;
185 for (i=0; i<kTotalVecCount; i++) // change this so we test all
186 {
187 err = clEnqueueNDRangeKernel( queue, kernel[i], 1, NULL, threads, NULL, 0, NULL, NULL );
188 if (err != CL_SUCCESS)
189 {
190 log_error("clEnqueueNDRangeKernel failed\n");
191 return -1;
192 }
193
194 err = clEnqueueReadBuffer( queue, streams[1], true, 0, sizeof(cl_float)*num_elements, (void *)output_ptr, 0, NULL, NULL );
195 if (err != CL_SUCCESS)
196 {
197 log_error("clEnqueueReadBuffer failed\n");
198 return -1;
199 }
200
201 if (verify_sign(input_ptr[0], output_ptr, n_elems*(i+1)))
202 {
203 log_error("%s test failed\n", fn_names[i]);
204 err = -1;
205 }
206 else
207 {
208 log_info("%s test passed\n", fn_names[i]);
209 err = 0;
210 }
211
212 if (err)
213 break;
214 }
215
216 clReleaseMemObject(streams[0]);
217 clReleaseMemObject(streams[1]);
218 for (i=0; i<kTotalVecCount; i++)
219 {
220 clReleaseKernel(kernel[i]);
221 clReleaseProgram(program[i]);
222 }
223 free(input_ptr[0]);
224 free(output_ptr);
225
226 if(err)
227 return err;
228
229 if( ! is_extension_available( device, "cl_khr_fp64"))
230 {
231 log_info( "skipping double test -- cl_khr_fp64 not supported.\n" );
232 return 0;
233 }
234
235 return test_sign_double( device, context, queue, n_elems);
236 }
237
238 #pragma mark -
239
240 const char *sign_kernel_code_double =
241 "#pragma OPENCL EXTENSION cl_khr_fp64 : enable\n"
242 "__kernel void test_sign_double(__global double *src, __global double *dst)\n"
243 "{\n"
244 " int tid = get_global_id(0);\n"
245 "\n"
246 " dst[tid] = sign(src[tid]);\n"
247 "}\n";
248
249 const char *sign2_kernel_code_double =
250 "#pragma OPENCL EXTENSION cl_khr_fp64 : enable\n"
251 "__kernel void test_sign2_double(__global double2 *src, __global double2 *dst)\n"
252 "{\n"
253 " int tid = get_global_id(0);\n"
254 "\n"
255 " dst[tid] = sign(src[tid]);\n"
256 "}\n";
257
258 const char *sign4_kernel_code_double =
259 "#pragma OPENCL EXTENSION cl_khr_fp64 : enable\n"
260 "__kernel void test_sign4_double(__global double4 *src, __global double4 *dst)\n"
261 "{\n"
262 " int tid = get_global_id(0);\n"
263 "\n"
264 " dst[tid] = sign(src[tid]);\n"
265 "}\n";
266
267 const char *sign8_kernel_code_double =
268 "#pragma OPENCL EXTENSION cl_khr_fp64 : enable\n"
269 "__kernel void test_sign8_double(__global double8 *src, __global double8 *dst)\n"
270 "{\n"
271 " int tid = get_global_id(0);\n"
272 "\n"
273 " dst[tid] = sign(src[tid]);\n"
274 "}\n";
275
276 const char *sign16_kernel_code_double =
277 "#pragma OPENCL EXTENSION cl_khr_fp64 : enable\n"
278 "__kernel void test_sign16_double(__global double16 *src, __global double16 *dst)\n"
279 "{\n"
280 " int tid = get_global_id(0);\n"
281 "\n"
282 " dst[tid] = sign(src[tid]);\n"
283 "}\n";
284
285 const char *sign3_kernel_code_double =
286 "#pragma OPENCL EXTENSION cl_khr_fp64 : enable\n"
287 "__kernel void test_sign3_double(__global double *src, __global double *dst)\n"
288 "{\n"
289 " int tid = get_global_id(0);\n"
290 "\n"
291 " vstore3(sign(vload3(tid,src)), tid, dst);\n"
292 "}\n";
293
294
295 static int
verify_sign_double(double * inptr,double * outptr,int n)296 verify_sign_double(double *inptr, double *outptr, int n)
297 {
298 double r;
299 int i;
300
301 for (i=0; i<n; i++)
302 {
303 if (inptr[i] > 0.0)
304 r = 1.0;
305 else if (inptr[i] < 0.0)
306 r = -1.0;
307 else
308 r = 0.0f;
309 if (r != outptr[i])
310 return -1;
311 }
312
313 return 0;
314 }
315
316 static const char *fn_names_double[] = { "SIGN double", "SIGN double2", "SIGN double4", "SIGN double8", "SIGN double16", "SIGN double3" };
317
318 int
test_sign_double(cl_device_id device,cl_context context,cl_command_queue queue,int n_elems)319 test_sign_double(cl_device_id device, cl_context context, cl_command_queue queue, int n_elems)
320 {
321 cl_mem streams[2];
322 cl_double *input_ptr[1], *output_ptr, *p;
323 cl_program program[kTotalVecCount];
324 cl_kernel kernel[kTotalVecCount];
325 void *values[2];
326 size_t threads[1];
327 int num_elements;
328 int err;
329 int i;
330 MTdata d;
331
332 num_elements = n_elems * 16;
333
334 input_ptr[0] = (cl_double*)malloc(sizeof(cl_double) * num_elements);
335 output_ptr = (cl_double*)malloc(sizeof(cl_double) * num_elements);
336 streams[0] = clCreateBuffer(context, CL_MEM_READ_WRITE,
337 sizeof(cl_double) * num_elements, NULL, NULL);
338 if (!streams[0])
339 {
340 log_error("clCreateBuffer failed\n");
341 return -1;
342 }
343
344 streams[1] = clCreateBuffer(context, CL_MEM_READ_WRITE,
345 sizeof(cl_double) * num_elements, NULL, NULL);
346 if (!streams[1])
347 {
348 log_error("clCreateBuffer failed\n");
349 return -1;
350 }
351
352 d = init_genrand( gRandomSeed );
353 p = input_ptr[0];
354 for (i=0; i<num_elements; i++)
355 p[i] = get_random_double(-0x20000000, 0x20000000, d);
356
357 free_mtdata(d); d = NULL;
358
359
360 err = clEnqueueWriteBuffer( queue, streams[0], true, 0, sizeof(cl_double)*num_elements, (void *)input_ptr[0], 0, NULL, NULL );
361 if (err != CL_SUCCESS)
362 {
363 log_error("clWriteArray failed\n");
364 return -1;
365 }
366
367 err = create_single_kernel_helper( context, &program[0], &kernel[0], 1, &sign_kernel_code_double, "test_sign_double" );
368 if (err)
369 return -1;
370 err = create_single_kernel_helper( context, &program[1], &kernel[1], 1, &sign2_kernel_code_double, "test_sign2_double" );
371 if (err)
372 return -1;
373 err = create_single_kernel_helper( context, &program[2], &kernel[2], 1, &sign4_kernel_code_double, "test_sign4_double" );
374 if (err)
375 return -1;
376 err = create_single_kernel_helper( context, &program[3], &kernel[3], 1, &sign8_kernel_code_double, "test_sign8_double" );
377 if (err)
378 return -1;
379 err = create_single_kernel_helper( context, &program[4], &kernel[4], 1, &sign16_kernel_code_double, "test_sign16_double" );
380 if (err)
381 return -1;
382 err = create_single_kernel_helper( context, &program[5], &kernel[5], 1, &sign3_kernel_code_double, "test_sign3_double" );
383 if (err)
384 return -1;
385
386 values[0] = streams[0];
387 values[1] = streams[1];
388 for (i=0; i<kTotalVecCount; i++)
389 {
390 err = clSetKernelArg(kernel[i], 0, sizeof streams[0], &streams[0] );
391 err |= clSetKernelArg(kernel[i], 1, sizeof streams[1], &streams[1] );
392 if (err != CL_SUCCESS)
393 {
394 log_error("clSetKernelArgs failed\n");
395 return -1;
396 }
397 }
398
399 threads[0] = (size_t)n_elems;
400 for (i=0; i<kTotalVecCount; i++) // this hsould be changed
401 {
402 err = clEnqueueNDRangeKernel( queue, kernel[i], 1, NULL, threads, NULL, 0, NULL, NULL );
403 if (err != CL_SUCCESS)
404 {
405 log_error("clEnqueueNDRangeKernel failed\n");
406 return -1;
407 }
408
409 err = clEnqueueReadBuffer( queue, streams[1], true, 0, sizeof(cl_double)*num_elements, (void *)output_ptr, 0, NULL, NULL );
410 if (err != CL_SUCCESS)
411 {
412 log_error("clEnqueueReadBuffer failed\n");
413 return -1;
414 }
415
416 if (verify_sign_double(input_ptr[0], output_ptr, n_elems*(i+1)))
417 {
418 log_error("%s test failed\n", fn_names_double[i]);
419 err = -1;
420 }
421 else
422 {
423 log_info("%s test passed\n", fn_names_double[i]);
424 err = 0;
425 }
426
427 if (err)
428 break;
429 }
430
431 clReleaseMemObject(streams[0]);
432 clReleaseMemObject(streams[1]);
433 for (i=0; i<kTotalVecCount; i++)
434 {
435 clReleaseKernel(kernel[i]);
436 clReleaseProgram(program[i]);
437 }
438 free(input_ptr[0]);
439 free(output_ptr);
440
441 return err;
442 }
443
444
445