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
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
|
//===----------------------------------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
///
/// \file
/// This file contains the implementation of the device kernels that wrap the
/// math functions from the hip-math provider.
///
//===----------------------------------------------------------------------===//
#ifdef HIP_MATH_FOUND
#include "Conformance/device_code/DeviceAPIs.hpp"
#include "Conformance/device_code/KernelRunner.hpp"
#include <gpuintrin.h>
#include <stddef.h>
using namespace kernels;
//===----------------------------------------------------------------------===//
// Helpers
//===----------------------------------------------------------------------===//
static inline float powfRoundedExponent(float Base, float Exponent) {
return __ocml_pow_f32(Base, __ocml_round_f32(Exponent));
}
static inline double sincosSin(double X) {
double CosX;
double SinX = __ocml_sincos_f64(X, &CosX);
return SinX;
}
static inline double sincosCos(double X) {
double CosX;
double SinX = __ocml_sincos_f64(X, &CosX);
return CosX;
}
static inline float sincosfSin(float X) {
float CosX;
float SinX = __ocml_sincos_f32(X, &CosX);
return SinX;
}
static inline float sincosfCos(float X) {
float CosX;
float SinX = __ocml_sincos_f32(X, &CosX);
return CosX;
}
//===----------------------------------------------------------------------===//
// Kernels
//===----------------------------------------------------------------------===//
extern "C" {
__gpu_kernel void acosKernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_acos_f64>(NumElements, Out, X);
}
__gpu_kernel void acosfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_acos_f32>(NumElements, Out, X);
}
__gpu_kernel void acosf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_acos_f16>(NumElements, Out, X);
}
__gpu_kernel void acoshfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_acosh_f32>(NumElements, Out, X);
}
__gpu_kernel void acoshf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_acosh_f16>(NumElements, Out, X);
}
__gpu_kernel void asinKernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_asin_f64>(NumElements, Out, X);
}
__gpu_kernel void asinfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_asin_f32>(NumElements, Out, X);
}
__gpu_kernel void asinf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_asin_f16>(NumElements, Out, X);
}
__gpu_kernel void asinhfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_asinh_f32>(NumElements, Out, X);
}
__gpu_kernel void asinhf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_asinh_f16>(NumElements, Out, X);
}
__gpu_kernel void atanfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_atan_f32>(NumElements, Out, X);
}
__gpu_kernel void atanf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_atan_f16>(NumElements, Out, X);
}
__gpu_kernel void atan2fKernel(const float *X, const float *Y, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_atan2_f32>(NumElements, Out, X, Y);
}
__gpu_kernel void atanhfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_atanh_f32>(NumElements, Out, X);
}
__gpu_kernel void atanhf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_atanh_f16>(NumElements, Out, X);
}
__gpu_kernel void cbrtKernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_cbrt_f64>(NumElements, Out, X);
}
__gpu_kernel void cbrtfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_cbrt_f32>(NumElements, Out, X);
}
__gpu_kernel void cosKernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_cos_f64>(NumElements, Out, X);
}
__gpu_kernel void cosfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_cos_f32>(NumElements, Out, X);
}
__gpu_kernel void cosf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_cos_f16>(NumElements, Out, X);
}
__gpu_kernel void coshfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_cosh_f32>(NumElements, Out, X);
}
__gpu_kernel void coshf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_cosh_f16>(NumElements, Out, X);
}
__gpu_kernel void cospifKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_cospi_f32>(NumElements, Out, X);
}
__gpu_kernel void erffKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_erf_f32>(NumElements, Out, X);
}
__gpu_kernel void expKernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_exp_f64>(NumElements, Out, X);
}
__gpu_kernel void expfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_exp_f32>(NumElements, Out, X);
}
__gpu_kernel void expf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_exp_f16>(NumElements, Out, X);
}
__gpu_kernel void exp10Kernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_exp10_f64>(NumElements, Out, X);
}
__gpu_kernel void exp10fKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_exp10_f32>(NumElements, Out, X);
}
__gpu_kernel void exp10f16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_exp10_f16>(NumElements, Out, X);
}
__gpu_kernel void exp2Kernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_exp2_f64>(NumElements, Out, X);
}
__gpu_kernel void exp2fKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_exp2_f32>(NumElements, Out, X);
}
__gpu_kernel void exp2f16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_exp2_f16>(NumElements, Out, X);
}
__gpu_kernel void expm1Kernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_expm1_f64>(NumElements, Out, X);
}
__gpu_kernel void expm1fKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_expm1_f32>(NumElements, Out, X);
}
__gpu_kernel void expm1f16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_expm1_f16>(NumElements, Out, X);
}
__gpu_kernel void hypotKernel(const double *X, const double *Y, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_hypot_f64>(NumElements, Out, X, Y);
}
__gpu_kernel void hypotfKernel(const float *X, const float *Y, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_hypot_f32>(NumElements, Out, X, Y);
}
__gpu_kernel void logKernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_log_f64>(NumElements, Out, X);
}
__gpu_kernel void logfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_log_f32>(NumElements, Out, X);
}
__gpu_kernel void logf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_log_f16>(NumElements, Out, X);
}
__gpu_kernel void log10Kernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_log10_f64>(NumElements, Out, X);
}
__gpu_kernel void log10fKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_log10_f32>(NumElements, Out, X);
}
__gpu_kernel void log10f16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_log10_f16>(NumElements, Out, X);
}
__gpu_kernel void log1pKernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_log1p_f64>(NumElements, Out, X);
}
__gpu_kernel void log1pfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_log1p_f32>(NumElements, Out, X);
}
__gpu_kernel void log2Kernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_log2_f64>(NumElements, Out, X);
}
__gpu_kernel void log2fKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_log2_f32>(NumElements, Out, X);
}
__gpu_kernel void log2f16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_log2_f16>(NumElements, Out, X);
}
__gpu_kernel void powfKernel(const float *X, float *Y, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_pow_f32>(NumElements, Out, X, Y);
}
__gpu_kernel void powfRoundedExponentKernel(const float *X, float *Y,
float *Out,
size_t NumElements) noexcept {
runKernelBody<powfRoundedExponent>(NumElements, Out, X, Y);
}
__gpu_kernel void sinKernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_sin_f64>(NumElements, Out, X);
}
__gpu_kernel void sinfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_sin_f32>(NumElements, Out, X);
}
__gpu_kernel void sinf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_sin_f16>(NumElements, Out, X);
}
__gpu_kernel void sincosSinKernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<sincosSin>(NumElements, Out, X);
}
__gpu_kernel void sincosCosKernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<sincosCos>(NumElements, Out, X);
}
__gpu_kernel void sincosfSinKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<sincosfSin>(NumElements, Out, X);
}
__gpu_kernel void sincosfCosKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<sincosfCos>(NumElements, Out, X);
}
__gpu_kernel void sinhfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_sinh_f32>(NumElements, Out, X);
}
__gpu_kernel void sinhf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_sinh_f16>(NumElements, Out, X);
}
__gpu_kernel void sinpifKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_sinpi_f32>(NumElements, Out, X);
}
__gpu_kernel void tanKernel(const double *X, double *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_tan_f64>(NumElements, Out, X);
}
__gpu_kernel void tanfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_tan_f32>(NumElements, Out, X);
}
__gpu_kernel void tanf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_tan_f16>(NumElements, Out, X);
}
__gpu_kernel void tanhfKernel(const float *X, float *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_tanh_f32>(NumElements, Out, X);
}
__gpu_kernel void tanhf16Kernel(const float16 *X, float16 *Out,
size_t NumElements) noexcept {
runKernelBody<__ocml_tanh_f16>(NumElements, Out, X);
}
} // extern "C"
#endif // HIP_MATH_FOUND
|
