Mercurial > hg > Gears > GearsAgda
annotate src/parallel_execution/helper_cuda.h @ 304:9755206813cb
helper_string.h for ANSI C
| author | Shinji KONO <kono@ie.u-ryukyu.ac.jp> |
|---|---|
| date | Tue, 14 Feb 2017 11:36:41 +0900 |
| parents | 1dbaef86593b |
| children | ec0a5b4fba05 |
| rev | line source |
|---|---|
| 291 | 1 /** |
| 2 * Copyright 1993-2013 NVIDIA Corporation. All rights reserved. | |
| 3 * | |
| 4 * Please refer to the NVIDIA end user license agreement (EULA) associated | |
| 5 * with this source code for terms and conditions that govern your use of | |
| 6 * this software. Any use, reproduction, disclosure, or distribution of | |
| 7 * this software and related documentation outside the terms of the EULA | |
| 8 * is strictly prohibited. | |
| 9 * | |
| 10 */ | |
| 11 | |
| 12 //////////////////////////////////////////////////////////////////////////////// | |
| 13 // These are CUDA Helper functions for initialization and error checking | |
| 14 | |
| 15 #ifndef HELPER_CUDA_H | |
| 16 #define HELPER_CUDA_H | |
| 17 | |
| 18 #pragma once | |
| 19 | |
| 20 #include <stdlib.h> | |
| 21 #include <stdio.h> | |
| 22 #include <string.h> | |
| 23 | |
| 24 #include "helper_string.h" | |
| 25 | |
| 26 #ifndef EXIT_WAIVED | |
| 27 #define EXIT_WAIVED 2 | |
| 28 #endif | |
| 29 | |
| 30 // Note, it is required that your SDK sample to include the proper header files, please | |
| 31 // refer the CUDA examples for examples of the needed CUDA headers, which may change depending | |
| 32 // on which CUDA functions are used. | |
| 33 | |
| 34 // CUDA Runtime error messages | |
| 35 #ifdef __DRIVER_TYPES_H__ | |
| 36 static const char *_cudaGetErrorEnum(cudaError_t error) | |
| 37 { | |
| 38 switch (error) | |
| 39 { | |
| 40 case cudaSuccess: | |
| 41 return "cudaSuccess"; | |
| 42 | |
| 43 case cudaErrorMissingConfiguration: | |
| 44 return "cudaErrorMissingConfiguration"; | |
| 45 | |
| 46 case cudaErrorMemoryAllocation: | |
| 47 return "cudaErrorMemoryAllocation"; | |
| 48 | |
| 49 case cudaErrorInitializationError: | |
| 50 return "cudaErrorInitializationError"; | |
| 51 | |
| 52 case cudaErrorLaunchFailure: | |
| 53 return "cudaErrorLaunchFailure"; | |
| 54 | |
| 55 case cudaErrorPriorLaunchFailure: | |
| 56 return "cudaErrorPriorLaunchFailure"; | |
| 57 | |
| 58 case cudaErrorLaunchTimeout: | |
| 59 return "cudaErrorLaunchTimeout"; | |
| 60 | |
| 61 case cudaErrorLaunchOutOfResources: | |
| 62 return "cudaErrorLaunchOutOfResources"; | |
| 63 | |
| 64 case cudaErrorInvalidDeviceFunction: | |
| 65 return "cudaErrorInvalidDeviceFunction"; | |
| 66 | |
| 67 case cudaErrorInvalidConfiguration: | |
| 68 return "cudaErrorInvalidConfiguration"; | |
| 69 | |
| 70 case cudaErrorInvalidDevice: | |
| 71 return "cudaErrorInvalidDevice"; | |
| 72 | |
| 73 case cudaErrorInvalidValue: | |
| 74 return "cudaErrorInvalidValue"; | |
| 75 | |
| 76 case cudaErrorInvalidPitchValue: | |
| 77 return "cudaErrorInvalidPitchValue"; | |
| 78 | |
| 79 case cudaErrorInvalidSymbol: | |
| 80 return "cudaErrorInvalidSymbol"; | |
| 81 | |
| 82 case cudaErrorMapBufferObjectFailed: | |
| 83 return "cudaErrorMapBufferObjectFailed"; | |
| 84 | |
| 85 case cudaErrorUnmapBufferObjectFailed: | |
| 86 return "cudaErrorUnmapBufferObjectFailed"; | |
| 87 | |
| 88 case cudaErrorInvalidHostPointer: | |
| 89 return "cudaErrorInvalidHostPointer"; | |
| 90 | |
| 91 case cudaErrorInvalidDevicePointer: | |
| 92 return "cudaErrorInvalidDevicePointer"; | |
| 93 | |
| 94 case cudaErrorInvalidTexture: | |
| 95 return "cudaErrorInvalidTexture"; | |
| 96 | |
| 97 case cudaErrorInvalidTextureBinding: | |
| 98 return "cudaErrorInvalidTextureBinding"; | |
| 99 | |
| 100 case cudaErrorInvalidChannelDescriptor: | |
| 101 return "cudaErrorInvalidChannelDescriptor"; | |
| 102 | |
| 103 case cudaErrorInvalidMemcpyDirection: | |
| 104 return "cudaErrorInvalidMemcpyDirection"; | |
| 105 | |
| 106 case cudaErrorAddressOfConstant: | |
| 107 return "cudaErrorAddressOfConstant"; | |
| 108 | |
| 109 case cudaErrorTextureFetchFailed: | |
| 110 return "cudaErrorTextureFetchFailed"; | |
| 111 | |
| 112 case cudaErrorTextureNotBound: | |
| 113 return "cudaErrorTextureNotBound"; | |
| 114 | |
| 115 case cudaErrorSynchronizationError: | |
| 116 return "cudaErrorSynchronizationError"; | |
| 117 | |
| 118 case cudaErrorInvalidFilterSetting: | |
| 119 return "cudaErrorInvalidFilterSetting"; | |
| 120 | |
| 121 case cudaErrorInvalidNormSetting: | |
| 122 return "cudaErrorInvalidNormSetting"; | |
| 123 | |
| 124 case cudaErrorMixedDeviceExecution: | |
| 125 return "cudaErrorMixedDeviceExecution"; | |
| 126 | |
| 127 case cudaErrorCudartUnloading: | |
| 128 return "cudaErrorCudartUnloading"; | |
| 129 | |
| 130 case cudaErrorUnknown: | |
| 131 return "cudaErrorUnknown"; | |
| 132 | |
| 133 case cudaErrorNotYetImplemented: | |
| 134 return "cudaErrorNotYetImplemented"; | |
| 135 | |
| 136 case cudaErrorMemoryValueTooLarge: | |
| 137 return "cudaErrorMemoryValueTooLarge"; | |
| 138 | |
| 139 case cudaErrorInvalidResourceHandle: | |
| 140 return "cudaErrorInvalidResourceHandle"; | |
| 141 | |
| 142 case cudaErrorNotReady: | |
| 143 return "cudaErrorNotReady"; | |
| 144 | |
| 145 case cudaErrorInsufficientDriver: | |
| 146 return "cudaErrorInsufficientDriver"; | |
| 147 | |
| 148 case cudaErrorSetOnActiveProcess: | |
| 149 return "cudaErrorSetOnActiveProcess"; | |
| 150 | |
| 151 case cudaErrorInvalidSurface: | |
| 152 return "cudaErrorInvalidSurface"; | |
| 153 | |
| 154 case cudaErrorNoDevice: | |
| 155 return "cudaErrorNoDevice"; | |
| 156 | |
| 157 case cudaErrorECCUncorrectable: | |
| 158 return "cudaErrorECCUncorrectable"; | |
| 159 | |
| 160 case cudaErrorSharedObjectSymbolNotFound: | |
| 161 return "cudaErrorSharedObjectSymbolNotFound"; | |
| 162 | |
| 163 case cudaErrorSharedObjectInitFailed: | |
| 164 return "cudaErrorSharedObjectInitFailed"; | |
| 165 | |
| 166 case cudaErrorUnsupportedLimit: | |
| 167 return "cudaErrorUnsupportedLimit"; | |
| 168 | |
| 169 case cudaErrorDuplicateVariableName: | |
| 170 return "cudaErrorDuplicateVariableName"; | |
| 171 | |
| 172 case cudaErrorDuplicateTextureName: | |
| 173 return "cudaErrorDuplicateTextureName"; | |
| 174 | |
| 175 case cudaErrorDuplicateSurfaceName: | |
| 176 return "cudaErrorDuplicateSurfaceName"; | |
| 177 | |
| 178 case cudaErrorDevicesUnavailable: | |
| 179 return "cudaErrorDevicesUnavailable"; | |
| 180 | |
| 181 case cudaErrorInvalidKernelImage: | |
| 182 return "cudaErrorInvalidKernelImage"; | |
| 183 | |
| 184 case cudaErrorNoKernelImageForDevice: | |
| 185 return "cudaErrorNoKernelImageForDevice"; | |
| 186 | |
| 187 case cudaErrorIncompatibleDriverContext: | |
| 188 return "cudaErrorIncompatibleDriverContext"; | |
| 189 | |
| 190 case cudaErrorPeerAccessAlreadyEnabled: | |
| 191 return "cudaErrorPeerAccessAlreadyEnabled"; | |
| 192 | |
| 193 case cudaErrorPeerAccessNotEnabled: | |
| 194 return "cudaErrorPeerAccessNotEnabled"; | |
| 195 | |
| 196 case cudaErrorDeviceAlreadyInUse: | |
| 197 return "cudaErrorDeviceAlreadyInUse"; | |
| 198 | |
| 199 case cudaErrorProfilerDisabled: | |
| 200 return "cudaErrorProfilerDisabled"; | |
| 201 | |
| 202 case cudaErrorProfilerNotInitialized: | |
| 203 return "cudaErrorProfilerNotInitialized"; | |
| 204 | |
| 205 case cudaErrorProfilerAlreadyStarted: | |
| 206 return "cudaErrorProfilerAlreadyStarted"; | |
| 207 | |
| 208 case cudaErrorProfilerAlreadyStopped: | |
| 209 return "cudaErrorProfilerAlreadyStopped"; | |
| 210 | |
| 211 /* Since CUDA 4.0*/ | |
| 212 case cudaErrorAssert: | |
| 213 return "cudaErrorAssert"; | |
| 214 | |
| 215 case cudaErrorTooManyPeers: | |
| 216 return "cudaErrorTooManyPeers"; | |
| 217 | |
| 218 case cudaErrorHostMemoryAlreadyRegistered: | |
| 219 return "cudaErrorHostMemoryAlreadyRegistered"; | |
| 220 | |
| 221 case cudaErrorHostMemoryNotRegistered: | |
| 222 return "cudaErrorHostMemoryNotRegistered"; | |
| 223 | |
| 224 /* Since CUDA 5.0 */ | |
| 225 case cudaErrorOperatingSystem: | |
| 226 return "cudaErrorOperatingSystem"; | |
| 227 | |
| 228 case cudaErrorPeerAccessUnsupported: | |
| 229 return "cudaErrorPeerAccessUnsupported"; | |
| 230 | |
| 231 case cudaErrorLaunchMaxDepthExceeded: | |
| 232 return "cudaErrorLaunchMaxDepthExceeded"; | |
| 233 | |
| 234 case cudaErrorLaunchFileScopedTex: | |
| 235 return "cudaErrorLaunchFileScopedTex"; | |
| 236 | |
| 237 case cudaErrorLaunchFileScopedSurf: | |
| 238 return "cudaErrorLaunchFileScopedSurf"; | |
| 239 | |
| 240 case cudaErrorSyncDepthExceeded: | |
| 241 return "cudaErrorSyncDepthExceeded"; | |
| 242 | |
| 243 case cudaErrorLaunchPendingCountExceeded: | |
| 244 return "cudaErrorLaunchPendingCountExceeded"; | |
| 245 | |
| 246 case cudaErrorNotPermitted: | |
| 247 return "cudaErrorNotPermitted"; | |
| 248 | |
| 249 case cudaErrorNotSupported: | |
| 250 return "cudaErrorNotSupported"; | |
| 251 | |
| 252 /* Since CUDA 6.0 */ | |
| 253 case cudaErrorHardwareStackError: | |
| 254 return "cudaErrorHardwareStackError"; | |
| 255 | |
| 256 case cudaErrorIllegalInstruction: | |
| 257 return "cudaErrorIllegalInstruction"; | |
| 258 | |
| 259 case cudaErrorMisalignedAddress: | |
| 260 return "cudaErrorMisalignedAddress"; | |
| 261 | |
| 262 case cudaErrorInvalidAddressSpace: | |
| 263 return "cudaErrorInvalidAddressSpace"; | |
| 264 | |
| 265 case cudaErrorInvalidPc: | |
| 266 return "cudaErrorInvalidPc"; | |
| 267 | |
| 268 case cudaErrorIllegalAddress: | |
| 269 return "cudaErrorIllegalAddress"; | |
| 270 | |
| 271 /* Since CUDA 6.5*/ | |
| 272 case cudaErrorInvalidPtx: | |
| 273 return "cudaErrorInvalidPtx"; | |
| 274 | |
| 275 case cudaErrorInvalidGraphicsContext: | |
| 276 return "cudaErrorInvalidGraphicsContext"; | |
| 277 | |
| 278 case cudaErrorStartupFailure: | |
| 279 return "cudaErrorStartupFailure"; | |
| 280 | |
| 281 case cudaErrorApiFailureBase: | |
| 282 return "cudaErrorApiFailureBase"; | |
| 283 | |
| 284 /* Since CUDA 8.0*/ | |
| 285 case cudaErrorNvlinkUncorrectable : | |
| 286 return "cudaErrorNvlinkUncorrectable"; | |
| 287 } | |
| 288 | |
| 289 return "<unknown>"; | |
| 290 } | |
|
304
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
291 #else |
| 291 | 292 // CUDA Driver API errors |
|
301
609bf62768b9
add -DUSE_CUDA=1 flag to cmake
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
291
diff
changeset
|
293 const char *_cudaGetErrorEnum(CUresult error) |
| 291 | 294 { |
| 295 switch (error) | |
| 296 { | |
| 297 case CUDA_SUCCESS: | |
| 298 return "CUDA_SUCCESS"; | |
| 299 | |
| 300 case CUDA_ERROR_INVALID_VALUE: | |
| 301 return "CUDA_ERROR_INVALID_VALUE"; | |
| 302 | |
| 303 case CUDA_ERROR_OUT_OF_MEMORY: | |
| 304 return "CUDA_ERROR_OUT_OF_MEMORY"; | |
| 305 | |
| 306 case CUDA_ERROR_NOT_INITIALIZED: | |
| 307 return "CUDA_ERROR_NOT_INITIALIZED"; | |
| 308 | |
| 309 case CUDA_ERROR_DEINITIALIZED: | |
| 310 return "CUDA_ERROR_DEINITIALIZED"; | |
| 311 | |
| 312 case CUDA_ERROR_PROFILER_DISABLED: | |
| 313 return "CUDA_ERROR_PROFILER_DISABLED"; | |
| 314 | |
| 315 case CUDA_ERROR_PROFILER_NOT_INITIALIZED: | |
| 316 return "CUDA_ERROR_PROFILER_NOT_INITIALIZED"; | |
| 317 | |
| 318 case CUDA_ERROR_PROFILER_ALREADY_STARTED: | |
| 319 return "CUDA_ERROR_PROFILER_ALREADY_STARTED"; | |
| 320 | |
| 321 case CUDA_ERROR_PROFILER_ALREADY_STOPPED: | |
| 322 return "CUDA_ERROR_PROFILER_ALREADY_STOPPED"; | |
| 323 | |
| 324 case CUDA_ERROR_NO_DEVICE: | |
| 325 return "CUDA_ERROR_NO_DEVICE"; | |
| 326 | |
| 327 case CUDA_ERROR_INVALID_DEVICE: | |
| 328 return "CUDA_ERROR_INVALID_DEVICE"; | |
| 329 | |
| 330 case CUDA_ERROR_INVALID_IMAGE: | |
| 331 return "CUDA_ERROR_INVALID_IMAGE"; | |
| 332 | |
| 333 case CUDA_ERROR_INVALID_CONTEXT: | |
| 334 return "CUDA_ERROR_INVALID_CONTEXT"; | |
| 335 | |
| 336 case CUDA_ERROR_CONTEXT_ALREADY_CURRENT: | |
| 337 return "CUDA_ERROR_CONTEXT_ALREADY_CURRENT"; | |
| 338 | |
| 339 case CUDA_ERROR_MAP_FAILED: | |
| 340 return "CUDA_ERROR_MAP_FAILED"; | |
| 341 | |
| 342 case CUDA_ERROR_UNMAP_FAILED: | |
| 343 return "CUDA_ERROR_UNMAP_FAILED"; | |
| 344 | |
| 345 case CUDA_ERROR_ARRAY_IS_MAPPED: | |
| 346 return "CUDA_ERROR_ARRAY_IS_MAPPED"; | |
| 347 | |
| 348 case CUDA_ERROR_ALREADY_MAPPED: | |
| 349 return "CUDA_ERROR_ALREADY_MAPPED"; | |
| 350 | |
| 351 case CUDA_ERROR_NO_BINARY_FOR_GPU: | |
| 352 return "CUDA_ERROR_NO_BINARY_FOR_GPU"; | |
| 353 | |
| 354 case CUDA_ERROR_ALREADY_ACQUIRED: | |
| 355 return "CUDA_ERROR_ALREADY_ACQUIRED"; | |
| 356 | |
| 357 case CUDA_ERROR_NOT_MAPPED: | |
| 358 return "CUDA_ERROR_NOT_MAPPED"; | |
| 359 | |
| 360 case CUDA_ERROR_NOT_MAPPED_AS_ARRAY: | |
| 361 return "CUDA_ERROR_NOT_MAPPED_AS_ARRAY"; | |
| 362 | |
| 363 case CUDA_ERROR_NOT_MAPPED_AS_POINTER: | |
| 364 return "CUDA_ERROR_NOT_MAPPED_AS_POINTER"; | |
| 365 | |
| 366 case CUDA_ERROR_ECC_UNCORRECTABLE: | |
| 367 return "CUDA_ERROR_ECC_UNCORRECTABLE"; | |
| 368 | |
| 369 case CUDA_ERROR_UNSUPPORTED_LIMIT: | |
| 370 return "CUDA_ERROR_UNSUPPORTED_LIMIT"; | |
| 371 | |
| 372 case CUDA_ERROR_CONTEXT_ALREADY_IN_USE: | |
| 373 return "CUDA_ERROR_CONTEXT_ALREADY_IN_USE"; | |
| 374 | |
| 375 case CUDA_ERROR_PEER_ACCESS_UNSUPPORTED: | |
| 376 return "CUDA_ERROR_PEER_ACCESS_UNSUPPORTED"; | |
| 377 | |
| 378 case CUDA_ERROR_INVALID_PTX: | |
| 379 return "CUDA_ERROR_INVALID_PTX"; | |
| 380 | |
| 381 case CUDA_ERROR_INVALID_GRAPHICS_CONTEXT: | |
| 382 return "CUDA_ERROR_INVALID_GRAPHICS_CONTEXT"; | |
| 383 | |
| 384 case CUDA_ERROR_NVLINK_UNCORRECTABLE: | |
| 385 return "CUDA_ERROR_NVLINK_UNCORRECTABLE"; | |
| 386 | |
| 387 case CUDA_ERROR_INVALID_SOURCE: | |
| 388 return "CUDA_ERROR_INVALID_SOURCE"; | |
| 389 | |
| 390 case CUDA_ERROR_FILE_NOT_FOUND: | |
| 391 return "CUDA_ERROR_FILE_NOT_FOUND"; | |
| 392 | |
| 393 case CUDA_ERROR_SHARED_OBJECT_SYMBOL_NOT_FOUND: | |
| 394 return "CUDA_ERROR_SHARED_OBJECT_SYMBOL_NOT_FOUND"; | |
| 395 | |
| 396 case CUDA_ERROR_SHARED_OBJECT_INIT_FAILED: | |
| 397 return "CUDA_ERROR_SHARED_OBJECT_INIT_FAILED"; | |
| 398 | |
| 399 case CUDA_ERROR_OPERATING_SYSTEM: | |
| 400 return "CUDA_ERROR_OPERATING_SYSTEM"; | |
| 401 | |
| 402 case CUDA_ERROR_INVALID_HANDLE: | |
| 403 return "CUDA_ERROR_INVALID_HANDLE"; | |
| 404 | |
| 405 case CUDA_ERROR_NOT_FOUND: | |
| 406 return "CUDA_ERROR_NOT_FOUND"; | |
| 407 | |
| 408 case CUDA_ERROR_NOT_READY: | |
| 409 return "CUDA_ERROR_NOT_READY"; | |
| 410 | |
| 411 case CUDA_ERROR_ILLEGAL_ADDRESS: | |
| 412 return "CUDA_ERROR_ILLEGAL_ADDRESS"; | |
| 413 | |
| 414 case CUDA_ERROR_LAUNCH_FAILED: | |
| 415 return "CUDA_ERROR_LAUNCH_FAILED"; | |
| 416 | |
| 417 case CUDA_ERROR_LAUNCH_OUT_OF_RESOURCES: | |
| 418 return "CUDA_ERROR_LAUNCH_OUT_OF_RESOURCES"; | |
| 419 | |
| 420 case CUDA_ERROR_LAUNCH_TIMEOUT: | |
| 421 return "CUDA_ERROR_LAUNCH_TIMEOUT"; | |
| 422 | |
| 423 case CUDA_ERROR_LAUNCH_INCOMPATIBLE_TEXTURING: | |
| 424 return "CUDA_ERROR_LAUNCH_INCOMPATIBLE_TEXTURING"; | |
| 425 | |
| 426 case CUDA_ERROR_PEER_ACCESS_ALREADY_ENABLED: | |
| 427 return "CUDA_ERROR_PEER_ACCESS_ALREADY_ENABLED"; | |
| 428 | |
| 429 case CUDA_ERROR_PEER_ACCESS_NOT_ENABLED: | |
| 430 return "CUDA_ERROR_PEER_ACCESS_NOT_ENABLED"; | |
| 431 | |
| 432 case CUDA_ERROR_PRIMARY_CONTEXT_ACTIVE: | |
| 433 return "CUDA_ERROR_PRIMARY_CONTEXT_ACTIVE"; | |
| 434 | |
| 435 case CUDA_ERROR_CONTEXT_IS_DESTROYED: | |
| 436 return "CUDA_ERROR_CONTEXT_IS_DESTROYED"; | |
| 437 | |
| 438 case CUDA_ERROR_ASSERT: | |
| 439 return "CUDA_ERROR_ASSERT"; | |
| 440 | |
| 441 case CUDA_ERROR_TOO_MANY_PEERS: | |
| 442 return "CUDA_ERROR_TOO_MANY_PEERS"; | |
| 443 | |
| 444 case CUDA_ERROR_HOST_MEMORY_ALREADY_REGISTERED: | |
| 445 return "CUDA_ERROR_HOST_MEMORY_ALREADY_REGISTERED"; | |
| 446 | |
| 447 case CUDA_ERROR_HOST_MEMORY_NOT_REGISTERED: | |
| 448 return "CUDA_ERROR_HOST_MEMORY_NOT_REGISTERED"; | |
| 449 | |
| 450 case CUDA_ERROR_HARDWARE_STACK_ERROR: | |
| 451 return "CUDA_ERROR_HARDWARE_STACK_ERROR"; | |
| 452 | |
| 453 case CUDA_ERROR_ILLEGAL_INSTRUCTION: | |
| 454 return "CUDA_ERROR_ILLEGAL_INSTRUCTION"; | |
| 455 | |
| 456 case CUDA_ERROR_MISALIGNED_ADDRESS: | |
| 457 return "CUDA_ERROR_MISALIGNED_ADDRESS"; | |
| 458 | |
| 459 case CUDA_ERROR_INVALID_ADDRESS_SPACE: | |
| 460 return "CUDA_ERROR_INVALID_ADDRESS_SPACE"; | |
| 461 | |
| 462 case CUDA_ERROR_INVALID_PC: | |
| 463 return "CUDA_ERROR_INVALID_PC"; | |
| 464 | |
| 465 case CUDA_ERROR_NOT_PERMITTED: | |
| 466 return "CUDA_ERROR_NOT_PERMITTED"; | |
| 467 | |
| 468 case CUDA_ERROR_NOT_SUPPORTED: | |
| 469 return "CUDA_ERROR_NOT_SUPPORTED"; | |
| 470 | |
| 471 case CUDA_ERROR_UNKNOWN: | |
| 472 return "CUDA_ERROR_UNKNOWN"; | |
| 473 } | |
| 474 | |
| 475 return "<unknown>"; | |
| 476 } | |
| 477 #endif | |
| 478 | |
| 479 #ifdef CUBLAS_API_H_ | |
| 480 // cuBLAS API errors | |
| 481 static const char *_cudaGetErrorEnum(cublasStatus_t error) | |
| 482 { | |
| 483 switch (error) | |
| 484 { | |
| 485 case CUBLAS_STATUS_SUCCESS: | |
| 486 return "CUBLAS_STATUS_SUCCESS"; | |
| 487 | |
| 488 case CUBLAS_STATUS_NOT_INITIALIZED: | |
| 489 return "CUBLAS_STATUS_NOT_INITIALIZED"; | |
| 490 | |
| 491 case CUBLAS_STATUS_ALLOC_FAILED: | |
| 492 return "CUBLAS_STATUS_ALLOC_FAILED"; | |
| 493 | |
| 494 case CUBLAS_STATUS_INVALID_VALUE: | |
| 495 return "CUBLAS_STATUS_INVALID_VALUE"; | |
| 496 | |
| 497 case CUBLAS_STATUS_ARCH_MISMATCH: | |
| 498 return "CUBLAS_STATUS_ARCH_MISMATCH"; | |
| 499 | |
| 500 case CUBLAS_STATUS_MAPPING_ERROR: | |
| 501 return "CUBLAS_STATUS_MAPPING_ERROR"; | |
| 502 | |
| 503 case CUBLAS_STATUS_EXECUTION_FAILED: | |
| 504 return "CUBLAS_STATUS_EXECUTION_FAILED"; | |
| 505 | |
| 506 case CUBLAS_STATUS_INTERNAL_ERROR: | |
| 507 return "CUBLAS_STATUS_INTERNAL_ERROR"; | |
| 508 | |
| 509 case CUBLAS_STATUS_NOT_SUPPORTED: | |
| 510 return "CUBLAS_STATUS_NOT_SUPPORTED"; | |
| 511 | |
| 512 case CUBLAS_STATUS_LICENSE_ERROR: | |
| 513 return "CUBLAS_STATUS_LICENSE_ERROR"; | |
| 514 } | |
| 515 | |
| 516 return "<unknown>"; | |
| 517 } | |
| 518 #endif | |
| 519 | |
| 520 #ifdef _CUFFT_H_ | |
| 521 // cuFFT API errors | |
| 522 static const char *_cudaGetErrorEnum(cufftResult error) | |
| 523 { | |
| 524 switch (error) | |
| 525 { | |
| 526 case CUFFT_SUCCESS: | |
| 527 return "CUFFT_SUCCESS"; | |
| 528 | |
| 529 case CUFFT_INVALID_PLAN: | |
| 530 return "CUFFT_INVALID_PLAN"; | |
| 531 | |
| 532 case CUFFT_ALLOC_FAILED: | |
| 533 return "CUFFT_ALLOC_FAILED"; | |
| 534 | |
| 535 case CUFFT_INVALID_TYPE: | |
| 536 return "CUFFT_INVALID_TYPE"; | |
| 537 | |
| 538 case CUFFT_INVALID_VALUE: | |
| 539 return "CUFFT_INVALID_VALUE"; | |
| 540 | |
| 541 case CUFFT_INTERNAL_ERROR: | |
| 542 return "CUFFT_INTERNAL_ERROR"; | |
| 543 | |
| 544 case CUFFT_EXEC_FAILED: | |
| 545 return "CUFFT_EXEC_FAILED"; | |
| 546 | |
| 547 case CUFFT_SETUP_FAILED: | |
| 548 return "CUFFT_SETUP_FAILED"; | |
| 549 | |
| 550 case CUFFT_INVALID_SIZE: | |
| 551 return "CUFFT_INVALID_SIZE"; | |
| 552 | |
| 553 case CUFFT_UNALIGNED_DATA: | |
| 554 return "CUFFT_UNALIGNED_DATA"; | |
| 555 | |
| 556 case CUFFT_INCOMPLETE_PARAMETER_LIST: | |
| 557 return "CUFFT_INCOMPLETE_PARAMETER_LIST"; | |
| 558 | |
| 559 case CUFFT_INVALID_DEVICE: | |
| 560 return "CUFFT_INVALID_DEVICE"; | |
| 561 | |
| 562 case CUFFT_PARSE_ERROR: | |
| 563 return "CUFFT_PARSE_ERROR"; | |
| 564 | |
| 565 case CUFFT_NO_WORKSPACE: | |
| 566 return "CUFFT_NO_WORKSPACE"; | |
| 567 | |
| 568 case CUFFT_NOT_IMPLEMENTED: | |
| 569 return "CUFFT_NOT_IMPLEMENTED"; | |
| 570 | |
| 571 case CUFFT_LICENSE_ERROR: | |
| 572 return "CUFFT_LICENSE_ERROR"; | |
| 573 | |
| 574 case CUFFT_NOT_SUPPORTED: | |
| 575 return "CUFFT_NOT_SUPPORTED"; | |
| 576 } | |
| 577 | |
| 578 return "<unknown>"; | |
| 579 } | |
| 580 #endif | |
| 581 | |
| 582 | |
| 583 #ifdef CUSPARSEAPI | |
| 584 // cuSPARSE API errors | |
| 585 static const char *_cudaGetErrorEnum(cusparseStatus_t error) | |
| 586 { | |
| 587 switch (error) | |
| 588 { | |
| 589 case CUSPARSE_STATUS_SUCCESS: | |
| 590 return "CUSPARSE_STATUS_SUCCESS"; | |
| 591 | |
| 592 case CUSPARSE_STATUS_NOT_INITIALIZED: | |
| 593 return "CUSPARSE_STATUS_NOT_INITIALIZED"; | |
| 594 | |
| 595 case CUSPARSE_STATUS_ALLOC_FAILED: | |
| 596 return "CUSPARSE_STATUS_ALLOC_FAILED"; | |
| 597 | |
| 598 case CUSPARSE_STATUS_INVALID_VALUE: | |
| 599 return "CUSPARSE_STATUS_INVALID_VALUE"; | |
| 600 | |
| 601 case CUSPARSE_STATUS_ARCH_MISMATCH: | |
| 602 return "CUSPARSE_STATUS_ARCH_MISMATCH"; | |
| 603 | |
| 604 case CUSPARSE_STATUS_MAPPING_ERROR: | |
| 605 return "CUSPARSE_STATUS_MAPPING_ERROR"; | |
| 606 | |
| 607 case CUSPARSE_STATUS_EXECUTION_FAILED: | |
| 608 return "CUSPARSE_STATUS_EXECUTION_FAILED"; | |
| 609 | |
| 610 case CUSPARSE_STATUS_INTERNAL_ERROR: | |
| 611 return "CUSPARSE_STATUS_INTERNAL_ERROR"; | |
| 612 | |
| 613 case CUSPARSE_STATUS_MATRIX_TYPE_NOT_SUPPORTED: | |
| 614 return "CUSPARSE_STATUS_MATRIX_TYPE_NOT_SUPPORTED"; | |
| 615 } | |
| 616 | |
| 617 return "<unknown>"; | |
| 618 } | |
| 619 #endif | |
| 620 | |
| 621 #ifdef CUSOLVER_COMMON_H_ | |
| 622 //cuSOLVER API errors | |
| 623 static const char *_cudaGetErrorEnum(cusolverStatus_t error) | |
| 624 { | |
| 625 switch(error) | |
| 626 { | |
| 627 case CUSOLVER_STATUS_SUCCESS: | |
| 628 return "CUSOLVER_STATUS_SUCCESS"; | |
| 629 case CUSOLVER_STATUS_NOT_INITIALIZED: | |
| 630 return "CUSOLVER_STATUS_NOT_INITIALIZED"; | |
| 631 case CUSOLVER_STATUS_ALLOC_FAILED: | |
| 632 return "CUSOLVER_STATUS_ALLOC_FAILED"; | |
| 633 case CUSOLVER_STATUS_INVALID_VALUE: | |
| 634 return "CUSOLVER_STATUS_INVALID_VALUE"; | |
| 635 case CUSOLVER_STATUS_ARCH_MISMATCH: | |
| 636 return "CUSOLVER_STATUS_ARCH_MISMATCH"; | |
| 637 case CUSOLVER_STATUS_MAPPING_ERROR: | |
| 638 return "CUSOLVER_STATUS_MAPPING_ERROR"; | |
| 639 case CUSOLVER_STATUS_EXECUTION_FAILED: | |
| 640 return "CUSOLVER_STATUS_EXECUTION_FAILED"; | |
| 641 case CUSOLVER_STATUS_INTERNAL_ERROR: | |
| 642 return "CUSOLVER_STATUS_INTERNAL_ERROR"; | |
| 643 case CUSOLVER_STATUS_MATRIX_TYPE_NOT_SUPPORTED: | |
| 644 return "CUSOLVER_STATUS_MATRIX_TYPE_NOT_SUPPORTED"; | |
| 645 case CUSOLVER_STATUS_NOT_SUPPORTED : | |
| 646 return "CUSOLVER_STATUS_NOT_SUPPORTED "; | |
| 647 case CUSOLVER_STATUS_ZERO_PIVOT: | |
| 648 return "CUSOLVER_STATUS_ZERO_PIVOT"; | |
| 649 case CUSOLVER_STATUS_INVALID_LICENSE: | |
| 650 return "CUSOLVER_STATUS_INVALID_LICENSE"; | |
| 651 } | |
| 652 | |
| 653 return "<unknown>"; | |
| 654 | |
| 655 } | |
| 656 #endif | |
| 657 | |
| 658 #ifdef CURAND_H_ | |
| 659 // cuRAND API errors | |
| 660 static const char *_cudaGetErrorEnum(curandStatus_t error) | |
| 661 { | |
| 662 switch (error) | |
| 663 { | |
| 664 case CURAND_STATUS_SUCCESS: | |
| 665 return "CURAND_STATUS_SUCCESS"; | |
| 666 | |
| 667 case CURAND_STATUS_VERSION_MISMATCH: | |
| 668 return "CURAND_STATUS_VERSION_MISMATCH"; | |
| 669 | |
| 670 case CURAND_STATUS_NOT_INITIALIZED: | |
| 671 return "CURAND_STATUS_NOT_INITIALIZED"; | |
| 672 | |
| 673 case CURAND_STATUS_ALLOCATION_FAILED: | |
| 674 return "CURAND_STATUS_ALLOCATION_FAILED"; | |
| 675 | |
| 676 case CURAND_STATUS_TYPE_ERROR: | |
| 677 return "CURAND_STATUS_TYPE_ERROR"; | |
| 678 | |
| 679 case CURAND_STATUS_OUT_OF_RANGE: | |
| 680 return "CURAND_STATUS_OUT_OF_RANGE"; | |
| 681 | |
| 682 case CURAND_STATUS_LENGTH_NOT_MULTIPLE: | |
| 683 return "CURAND_STATUS_LENGTH_NOT_MULTIPLE"; | |
| 684 | |
| 685 case CURAND_STATUS_DOUBLE_PRECISION_REQUIRED: | |
| 686 return "CURAND_STATUS_DOUBLE_PRECISION_REQUIRED"; | |
| 687 | |
| 688 case CURAND_STATUS_LAUNCH_FAILURE: | |
| 689 return "CURAND_STATUS_LAUNCH_FAILURE"; | |
| 690 | |
| 691 case CURAND_STATUS_PREEXISTING_FAILURE: | |
| 692 return "CURAND_STATUS_PREEXISTING_FAILURE"; | |
| 693 | |
| 694 case CURAND_STATUS_INITIALIZATION_FAILED: | |
| 695 return "CURAND_STATUS_INITIALIZATION_FAILED"; | |
| 696 | |
| 697 case CURAND_STATUS_ARCH_MISMATCH: | |
| 698 return "CURAND_STATUS_ARCH_MISMATCH"; | |
| 699 | |
| 700 case CURAND_STATUS_INTERNAL_ERROR: | |
| 701 return "CURAND_STATUS_INTERNAL_ERROR"; | |
| 702 } | |
| 703 | |
| 704 return "<unknown>"; | |
| 705 } | |
| 706 #endif | |
| 707 | |
| 708 #ifdef NV_NPPIDEFS_H | |
| 709 // NPP API errors | |
| 710 static const char *_cudaGetErrorEnum(NppStatus error) | |
| 711 { | |
| 712 switch (error) | |
| 713 { | |
| 714 case NPP_NOT_SUPPORTED_MODE_ERROR: | |
| 715 return "NPP_NOT_SUPPORTED_MODE_ERROR"; | |
| 716 | |
| 717 case NPP_ROUND_MODE_NOT_SUPPORTED_ERROR: | |
| 718 return "NPP_ROUND_MODE_NOT_SUPPORTED_ERROR"; | |
| 719 | |
| 720 case NPP_RESIZE_NO_OPERATION_ERROR: | |
| 721 return "NPP_RESIZE_NO_OPERATION_ERROR"; | |
| 722 | |
| 723 case NPP_NOT_SUFFICIENT_COMPUTE_CAPABILITY: | |
| 724 return "NPP_NOT_SUFFICIENT_COMPUTE_CAPABILITY"; | |
| 725 | |
| 726 #if ((NPP_VERSION_MAJOR << 12) + (NPP_VERSION_MINOR << 4)) <= 0x5000 | |
| 727 | |
| 728 case NPP_BAD_ARG_ERROR: | |
| 729 return "NPP_BAD_ARGUMENT_ERROR"; | |
| 730 | |
| 731 case NPP_COEFF_ERROR: | |
| 732 return "NPP_COEFFICIENT_ERROR"; | |
| 733 | |
| 734 case NPP_RECT_ERROR: | |
| 735 return "NPP_RECTANGLE_ERROR"; | |
| 736 | |
| 737 case NPP_QUAD_ERROR: | |
| 738 return "NPP_QUADRANGLE_ERROR"; | |
| 739 | |
| 740 case NPP_MEM_ALLOC_ERR: | |
| 741 return "NPP_MEMORY_ALLOCATION_ERROR"; | |
| 742 | |
| 743 case NPP_HISTO_NUMBER_OF_LEVELS_ERROR: | |
| 744 return "NPP_HISTOGRAM_NUMBER_OF_LEVELS_ERROR"; | |
| 745 | |
| 746 case NPP_INVALID_INPUT: | |
| 747 return "NPP_INVALID_INPUT"; | |
| 748 | |
| 749 case NPP_POINTER_ERROR: | |
| 750 return "NPP_POINTER_ERROR"; | |
| 751 | |
| 752 case NPP_WARNING: | |
| 753 return "NPP_WARNING"; | |
| 754 | |
| 755 case NPP_ODD_ROI_WARNING: | |
| 756 return "NPP_ODD_ROI_WARNING"; | |
| 757 #else | |
| 758 | |
| 759 // These are for CUDA 5.5 or higher | |
| 760 case NPP_BAD_ARGUMENT_ERROR: | |
| 761 return "NPP_BAD_ARGUMENT_ERROR"; | |
| 762 | |
| 763 case NPP_COEFFICIENT_ERROR: | |
| 764 return "NPP_COEFFICIENT_ERROR"; | |
| 765 | |
| 766 case NPP_RECTANGLE_ERROR: | |
| 767 return "NPP_RECTANGLE_ERROR"; | |
| 768 | |
| 769 case NPP_QUADRANGLE_ERROR: | |
| 770 return "NPP_QUADRANGLE_ERROR"; | |
| 771 | |
| 772 case NPP_MEMORY_ALLOCATION_ERR: | |
| 773 return "NPP_MEMORY_ALLOCATION_ERROR"; | |
| 774 | |
| 775 case NPP_HISTOGRAM_NUMBER_OF_LEVELS_ERROR: | |
| 776 return "NPP_HISTOGRAM_NUMBER_OF_LEVELS_ERROR"; | |
| 777 | |
| 778 case NPP_INVALID_HOST_POINTER_ERROR: | |
| 779 return "NPP_INVALID_HOST_POINTER_ERROR"; | |
| 780 | |
| 781 case NPP_INVALID_DEVICE_POINTER_ERROR: | |
| 782 return "NPP_INVALID_DEVICE_POINTER_ERROR"; | |
| 783 #endif | |
| 784 | |
| 785 case NPP_LUT_NUMBER_OF_LEVELS_ERROR: | |
| 786 return "NPP_LUT_NUMBER_OF_LEVELS_ERROR"; | |
| 787 | |
| 788 case NPP_TEXTURE_BIND_ERROR: | |
| 789 return "NPP_TEXTURE_BIND_ERROR"; | |
| 790 | |
| 791 case NPP_WRONG_INTERSECTION_ROI_ERROR: | |
| 792 return "NPP_WRONG_INTERSECTION_ROI_ERROR"; | |
| 793 | |
| 794 case NPP_NOT_EVEN_STEP_ERROR: | |
| 795 return "NPP_NOT_EVEN_STEP_ERROR"; | |
| 796 | |
| 797 case NPP_INTERPOLATION_ERROR: | |
| 798 return "NPP_INTERPOLATION_ERROR"; | |
| 799 | |
| 800 case NPP_RESIZE_FACTOR_ERROR: | |
| 801 return "NPP_RESIZE_FACTOR_ERROR"; | |
| 802 | |
| 803 case NPP_HAAR_CLASSIFIER_PIXEL_MATCH_ERROR: | |
| 804 return "NPP_HAAR_CLASSIFIER_PIXEL_MATCH_ERROR"; | |
| 805 | |
| 806 | |
| 807 #if ((NPP_VERSION_MAJOR << 12) + (NPP_VERSION_MINOR << 4)) <= 0x5000 | |
| 808 | |
| 809 case NPP_MEMFREE_ERR: | |
| 810 return "NPP_MEMFREE_ERR"; | |
| 811 | |
| 812 case NPP_MEMSET_ERR: | |
| 813 return "NPP_MEMSET_ERR"; | |
| 814 | |
| 815 case NPP_MEMCPY_ERR: | |
| 816 return "NPP_MEMCPY_ERROR"; | |
| 817 | |
| 818 case NPP_MIRROR_FLIP_ERR: | |
| 819 return "NPP_MIRROR_FLIP_ERR"; | |
| 820 #else | |
| 821 | |
| 822 case NPP_MEMFREE_ERROR: | |
| 823 return "NPP_MEMFREE_ERROR"; | |
| 824 | |
| 825 case NPP_MEMSET_ERROR: | |
| 826 return "NPP_MEMSET_ERROR"; | |
| 827 | |
| 828 case NPP_MEMCPY_ERROR: | |
| 829 return "NPP_MEMCPY_ERROR"; | |
| 830 | |
| 831 case NPP_MIRROR_FLIP_ERROR: | |
| 832 return "NPP_MIRROR_FLIP_ERROR"; | |
| 833 #endif | |
| 834 | |
| 835 case NPP_ALIGNMENT_ERROR: | |
| 836 return "NPP_ALIGNMENT_ERROR"; | |
| 837 | |
| 838 case NPP_STEP_ERROR: | |
| 839 return "NPP_STEP_ERROR"; | |
| 840 | |
| 841 case NPP_SIZE_ERROR: | |
| 842 return "NPP_SIZE_ERROR"; | |
| 843 | |
| 844 case NPP_NULL_POINTER_ERROR: | |
| 845 return "NPP_NULL_POINTER_ERROR"; | |
| 846 | |
| 847 case NPP_CUDA_KERNEL_EXECUTION_ERROR: | |
| 848 return "NPP_CUDA_KERNEL_EXECUTION_ERROR"; | |
| 849 | |
| 850 case NPP_NOT_IMPLEMENTED_ERROR: | |
| 851 return "NPP_NOT_IMPLEMENTED_ERROR"; | |
| 852 | |
| 853 case NPP_ERROR: | |
| 854 return "NPP_ERROR"; | |
| 855 | |
| 856 case NPP_SUCCESS: | |
| 857 return "NPP_SUCCESS"; | |
| 858 | |
| 859 case NPP_WRONG_INTERSECTION_QUAD_WARNING: | |
| 860 return "NPP_WRONG_INTERSECTION_QUAD_WARNING"; | |
| 861 | |
| 862 case NPP_MISALIGNED_DST_ROI_WARNING: | |
| 863 return "NPP_MISALIGNED_DST_ROI_WARNING"; | |
| 864 | |
| 865 case NPP_AFFINE_QUAD_INCORRECT_WARNING: | |
| 866 return "NPP_AFFINE_QUAD_INCORRECT_WARNING"; | |
| 867 | |
| 868 case NPP_DOUBLE_SIZE_WARNING: | |
| 869 return "NPP_DOUBLE_SIZE_WARNING"; | |
| 870 | |
| 871 case NPP_WRONG_INTERSECTION_ROI_WARNING: | |
| 872 return "NPP_WRONG_INTERSECTION_ROI_WARNING"; | |
| 873 | |
| 874 #if ((NPP_VERSION_MAJOR << 12) + (NPP_VERSION_MINOR << 4)) >= 0x6000 | |
| 875 /* These are 6.0 or higher */ | |
| 876 case NPP_LUT_PALETTE_BITSIZE_ERROR: | |
| 877 return "NPP_LUT_PALETTE_BITSIZE_ERROR"; | |
| 878 | |
| 879 case NPP_ZC_MODE_NOT_SUPPORTED_ERROR: | |
| 880 return "NPP_ZC_MODE_NOT_SUPPORTED_ERROR"; | |
| 881 | |
| 882 case NPP_QUALITY_INDEX_ERROR: | |
| 883 return "NPP_QUALITY_INDEX_ERROR"; | |
| 884 | |
| 885 case NPP_CHANNEL_ORDER_ERROR: | |
| 886 return "NPP_CHANNEL_ORDER_ERROR"; | |
| 887 | |
| 888 case NPP_ZERO_MASK_VALUE_ERROR: | |
| 889 return "NPP_ZERO_MASK_VALUE_ERROR"; | |
| 890 | |
| 891 case NPP_NUMBER_OF_CHANNELS_ERROR: | |
| 892 return "NPP_NUMBER_OF_CHANNELS_ERROR"; | |
| 893 | |
| 894 case NPP_COI_ERROR: | |
| 895 return "NPP_COI_ERROR"; | |
| 896 | |
| 897 case NPP_DIVISOR_ERROR: | |
| 898 return "NPP_DIVISOR_ERROR"; | |
| 899 | |
| 900 case NPP_CHANNEL_ERROR: | |
| 901 return "NPP_CHANNEL_ERROR"; | |
| 902 | |
| 903 case NPP_STRIDE_ERROR: | |
| 904 return "NPP_STRIDE_ERROR"; | |
| 905 | |
| 906 case NPP_ANCHOR_ERROR: | |
| 907 return "NPP_ANCHOR_ERROR"; | |
| 908 | |
| 909 case NPP_MASK_SIZE_ERROR: | |
| 910 return "NPP_MASK_SIZE_ERROR"; | |
| 911 | |
| 912 case NPP_MOMENT_00_ZERO_ERROR: | |
| 913 return "NPP_MOMENT_00_ZERO_ERROR"; | |
| 914 | |
| 915 case NPP_THRESHOLD_NEGATIVE_LEVEL_ERROR: | |
| 916 return "NPP_THRESHOLD_NEGATIVE_LEVEL_ERROR"; | |
| 917 | |
| 918 case NPP_THRESHOLD_ERROR: | |
| 919 return "NPP_THRESHOLD_ERROR"; | |
| 920 | |
| 921 case NPP_CONTEXT_MATCH_ERROR: | |
| 922 return "NPP_CONTEXT_MATCH_ERROR"; | |
| 923 | |
| 924 case NPP_FFT_FLAG_ERROR: | |
| 925 return "NPP_FFT_FLAG_ERROR"; | |
| 926 | |
| 927 case NPP_FFT_ORDER_ERROR: | |
| 928 return "NPP_FFT_ORDER_ERROR"; | |
| 929 | |
| 930 case NPP_SCALE_RANGE_ERROR: | |
| 931 return "NPP_SCALE_RANGE_ERROR"; | |
| 932 | |
| 933 case NPP_DATA_TYPE_ERROR: | |
| 934 return "NPP_DATA_TYPE_ERROR"; | |
| 935 | |
| 936 case NPP_OUT_OFF_RANGE_ERROR: | |
| 937 return "NPP_OUT_OFF_RANGE_ERROR"; | |
| 938 | |
| 939 case NPP_DIVIDE_BY_ZERO_ERROR: | |
| 940 return "NPP_DIVIDE_BY_ZERO_ERROR"; | |
| 941 | |
| 942 case NPP_RANGE_ERROR: | |
| 943 return "NPP_RANGE_ERROR"; | |
| 944 | |
| 945 case NPP_NO_MEMORY_ERROR: | |
| 946 return "NPP_NO_MEMORY_ERROR"; | |
| 947 | |
| 948 case NPP_ERROR_RESERVED: | |
| 949 return "NPP_ERROR_RESERVED"; | |
| 950 | |
| 951 case NPP_NO_OPERATION_WARNING: | |
| 952 return "NPP_NO_OPERATION_WARNING"; | |
| 953 | |
| 954 case NPP_DIVIDE_BY_ZERO_WARNING: | |
| 955 return "NPP_DIVIDE_BY_ZERO_WARNING"; | |
| 956 #endif | |
| 957 | |
| 958 #if ((NPP_VERSION_MAJOR << 12) + (NPP_VERSION_MINOR << 4)) >= 0x7000 | |
| 959 /* These are 7.0 or higher */ | |
| 960 case NPP_OVERFLOW_ERROR: | |
| 961 return "NPP_OVERFLOW_ERROR"; | |
| 962 | |
| 963 case NPP_CORRUPTED_DATA_ERROR: | |
| 964 return "NPP_CORRUPTED_DATA_ERROR"; | |
| 965 #endif | |
| 966 } | |
| 967 | |
| 968 return "<unknown>"; | |
| 969 } | |
| 970 #endif | |
| 971 | |
| 972 #ifdef __DRIVER_TYPES_H__ | |
| 973 #ifndef DEVICE_RESET | |
| 974 #define DEVICE_RESET cudaDeviceReset(); | |
| 975 #endif | |
| 976 #else | |
| 977 #ifndef DEVICE_RESET | |
| 978 #define DEVICE_RESET | |
| 979 #endif | |
| 980 #endif | |
| 981 | |
|
304
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
982 #ifndef __DRIVER_TYPES_H__ |
|
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
983 static inline void check(CUresult result, char const *const func, const char *const file, int const line) |
| 291 | 984 { |
| 985 if (result) | |
| 986 { | |
| 987 fprintf(stderr, "CUDA error at %s:%d code=%d(%s) \"%s\" \n", | |
|
304
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
988 file, line, (unsigned int)(result), _cudaGetErrorEnum(result), func); |
| 291 | 989 DEVICE_RESET |
| 990 // Make sure we call CUDA Device Reset before exiting | |
| 991 exit(EXIT_FAILURE); | |
| 992 } | |
| 993 } | |
|
304
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
994 #else |
|
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
995 static inline void check(cudaError_t result, char const *const func, const char *const file, int const line) |
|
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
996 { |
|
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
997 if (result) |
|
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
998 { |
|
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
999 fprintf(stderr, "CUDA error at %s:%d code=%d(%s) \"%s\" \n", |
|
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
1000 file, line, (unsigned int)(result), _cudaGetErrorEnum(result), func); |
|
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
1001 DEVICE_RESET |
|
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
1002 // Make sure we call CUDA Device Reset before exiting |
|
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
1003 exit(EXIT_FAILURE); |
|
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
1004 } |
|
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
1005 } |
|
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
1006 #endif |
| 291 | 1007 |
| 1008 #ifdef __DRIVER_TYPES_H__ | |
| 1009 // This will output the proper CUDA error strings in the event that a CUDA host call returns an error | |
| 1010 #define checkCudaErrors(val) check ( (val), #val, __FILE__, __LINE__ ) | |
| 1011 | |
| 1012 // This will output the proper error string when calling cudaGetLastError | |
| 1013 #define getLastCudaError(msg) __getLastCudaError (msg, __FILE__, __LINE__) | |
| 1014 | |
| 1015 inline void __getLastCudaError(const char *errorMessage, const char *file, const int line) | |
| 1016 { | |
| 1017 cudaError_t err = cudaGetLastError(); | |
| 1018 | |
| 1019 if (cudaSuccess != err) | |
| 1020 { | |
| 1021 fprintf(stderr, "%s(%i) : getLastCudaError() CUDA error : %s : (%d) %s.\n", | |
| 1022 file, line, errorMessage, (int)err, cudaGetErrorString(err)); | |
| 1023 DEVICE_RESET | |
| 1024 exit(EXIT_FAILURE); | |
| 1025 } | |
| 1026 } | |
| 1027 #endif | |
| 1028 | |
| 1029 #ifndef MAX | |
| 1030 #define MAX(a,b) (a > b ? a : b) | |
| 1031 #endif | |
| 1032 | |
| 1033 // Float To Int conversion | |
| 1034 inline int ftoi(float value) | |
| 1035 { | |
| 1036 return (value >= 0 ? (int)(value + 0.5) : (int)(value - 0.5)); | |
| 1037 } | |
| 1038 | |
| 1039 // Beginning of GPU Architecture definitions | |
| 1040 inline int _ConvertSMVer2Cores(int major, int minor) | |
| 1041 { | |
| 1042 // Defines for GPU Architecture types (using the SM version to determine the # of cores per SM | |
|
304
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
1043 typedef struct sSMtoCores |
| 291 | 1044 { |
| 1045 int SM; // 0xMm (hexidecimal notation), M = SM Major version, and m = SM minor version | |
| 1046 int Cores; | |
| 1047 } sSMtoCores; | |
| 1048 | |
| 1049 sSMtoCores nGpuArchCoresPerSM[] = | |
| 1050 { | |
| 1051 { 0x20, 32 }, // Fermi Generation (SM 2.0) GF100 class | |
| 1052 { 0x21, 48 }, // Fermi Generation (SM 2.1) GF10x class | |
| 1053 { 0x30, 192}, // Kepler Generation (SM 3.0) GK10x class | |
| 1054 { 0x32, 192}, // Kepler Generation (SM 3.2) GK10x class | |
| 1055 { 0x35, 192}, // Kepler Generation (SM 3.5) GK11x class | |
| 1056 { 0x37, 192}, // Kepler Generation (SM 3.7) GK21x class | |
| 1057 { 0x50, 128}, // Maxwell Generation (SM 5.0) GM10x class | |
| 1058 { 0x52, 128}, // Maxwell Generation (SM 5.2) GM20x class | |
| 1059 { 0x53, 128}, // Maxwell Generation (SM 5.3) GM20x class | |
| 1060 { 0x60, 64 }, // Pascal Generation (SM 6.0) GP100 class | |
| 1061 { 0x61, 128}, // Pascal Generation (SM 6.1) GP10x class | |
| 1062 { 0x62, 128}, // Pascal Generation (SM 6.2) GP10x class | |
| 1063 { -1, -1 } | |
| 1064 }; | |
| 1065 | |
| 1066 int index = 0; | |
| 1067 | |
| 1068 while (nGpuArchCoresPerSM[index].SM != -1) | |
| 1069 { | |
| 1070 if (nGpuArchCoresPerSM[index].SM == ((major << 4) + minor)) | |
| 1071 { | |
| 1072 return nGpuArchCoresPerSM[index].Cores; | |
| 1073 } | |
| 1074 | |
| 1075 index++; | |
| 1076 } | |
| 1077 | |
| 1078 // If we don't find the values, we default use the previous one to run properly | |
| 1079 printf("MapSMtoCores for SM %d.%d is undefined. Default to use %d Cores/SM\n", major, minor, nGpuArchCoresPerSM[index-1].Cores); | |
| 1080 return nGpuArchCoresPerSM[index-1].Cores; | |
| 1081 } | |
| 1082 // end of GPU Architecture definitions | |
| 1083 | |
| 1084 #ifdef __CUDA_RUNTIME_H__ | |
| 1085 // General GPU Device CUDA Initialization | |
| 1086 inline int gpuDeviceInit(int devID) | |
| 1087 { | |
| 1088 int device_count; | |
| 1089 checkCudaErrors(cudaGetDeviceCount(&device_count)); | |
| 1090 | |
| 1091 if (device_count == 0) | |
| 1092 { | |
| 1093 fprintf(stderr, "gpuDeviceInit() CUDA error: no devices supporting CUDA.\n"); | |
| 1094 exit(EXIT_FAILURE); | |
| 1095 } | |
| 1096 | |
| 1097 if (devID < 0) | |
| 1098 { | |
| 1099 devID = 0; | |
| 1100 } | |
| 1101 | |
| 1102 if (devID > device_count-1) | |
| 1103 { | |
| 1104 fprintf(stderr, "\n"); | |
| 1105 fprintf(stderr, ">> %d CUDA capable GPU device(s) detected. <<\n", device_count); | |
| 1106 fprintf(stderr, ">> gpuDeviceInit (-device=%d) is not a valid GPU device. <<\n", devID); | |
| 1107 fprintf(stderr, "\n"); | |
| 1108 return -devID; | |
| 1109 } | |
| 1110 | |
|
304
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
1111 struct cudaDeviceProp deviceProp; |
| 291 | 1112 checkCudaErrors(cudaGetDeviceProperties(&deviceProp, devID)); |
| 1113 | |
| 1114 if (deviceProp.computeMode == cudaComputeModeProhibited) | |
| 1115 { | |
| 1116 fprintf(stderr, "Error: device is running in <Compute Mode Prohibited>, no threads can use ::cudaSetDevice().\n"); | |
| 1117 return -1; | |
| 1118 } | |
| 1119 | |
| 1120 if (deviceProp.major < 1) | |
| 1121 { | |
| 1122 fprintf(stderr, "gpuDeviceInit(): GPU device does not support CUDA.\n"); | |
| 1123 exit(EXIT_FAILURE); | |
| 1124 } | |
| 1125 | |
| 1126 checkCudaErrors(cudaSetDevice(devID)); | |
| 1127 printf("gpuDeviceInit() CUDA Device [%d]: \"%s\n", devID, deviceProp.name); | |
| 1128 | |
| 1129 return devID; | |
| 1130 } | |
| 1131 | |
| 1132 // This function returns the best GPU (with maximum GFLOPS) | |
| 1133 inline int gpuGetMaxGflopsDeviceId() | |
| 1134 { | |
| 1135 int current_device = 0, sm_per_multiproc = 0; | |
| 1136 int max_perf_device = 0; | |
| 1137 int device_count = 0, best_SM_arch = 0; | |
| 1138 int devices_prohibited = 0; | |
| 1139 | |
| 1140 unsigned long long max_compute_perf = 0; | |
|
304
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
1141 struct cudaDeviceProp deviceProp; |
| 291 | 1142 cudaGetDeviceCount(&device_count); |
| 1143 | |
| 1144 checkCudaErrors(cudaGetDeviceCount(&device_count)); | |
| 1145 | |
| 1146 if (device_count == 0) | |
| 1147 { | |
| 1148 fprintf(stderr, "gpuGetMaxGflopsDeviceId() CUDA error: no devices supporting CUDA.\n"); | |
| 1149 exit(EXIT_FAILURE); | |
| 1150 } | |
| 1151 | |
| 1152 // Find the best major SM Architecture GPU device | |
| 1153 while (current_device < device_count) | |
| 1154 { | |
| 1155 cudaGetDeviceProperties(&deviceProp, current_device); | |
| 1156 | |
| 1157 // If this GPU is not running on Compute Mode prohibited, then we can add it to the list | |
| 1158 if (deviceProp.computeMode != cudaComputeModeProhibited) | |
| 1159 { | |
| 1160 if (deviceProp.major > 0 && deviceProp.major < 9999) | |
| 1161 { | |
| 1162 best_SM_arch = MAX(best_SM_arch, deviceProp.major); | |
| 1163 } | |
| 1164 } | |
| 1165 else | |
| 1166 { | |
| 1167 devices_prohibited++; | |
| 1168 } | |
| 1169 | |
| 1170 current_device++; | |
| 1171 } | |
| 1172 | |
| 1173 if (devices_prohibited == device_count) | |
| 1174 { | |
| 1175 fprintf(stderr, "gpuGetMaxGflopsDeviceId() CUDA error: all devices have compute mode prohibited.\n"); | |
| 1176 exit(EXIT_FAILURE); | |
| 1177 } | |
| 1178 | |
| 1179 // Find the best CUDA capable GPU device | |
| 1180 current_device = 0; | |
| 1181 | |
| 1182 while (current_device < device_count) | |
| 1183 { | |
| 1184 cudaGetDeviceProperties(&deviceProp, current_device); | |
| 1185 | |
| 1186 // If this GPU is not running on Compute Mode prohibited, then we can add it to the list | |
| 1187 if (deviceProp.computeMode != cudaComputeModeProhibited) | |
| 1188 { | |
| 1189 if (deviceProp.major == 9999 && deviceProp.minor == 9999) | |
| 1190 { | |
| 1191 sm_per_multiproc = 1; | |
| 1192 } | |
| 1193 else | |
| 1194 { | |
| 1195 sm_per_multiproc = _ConvertSMVer2Cores(deviceProp.major, deviceProp.minor); | |
| 1196 } | |
| 1197 | |
| 1198 unsigned long long compute_perf = (unsigned long long) deviceProp.multiProcessorCount * sm_per_multiproc * deviceProp.clockRate; | |
| 1199 | |
| 1200 if (compute_perf > max_compute_perf) | |
| 1201 { | |
| 1202 // If we find GPU with SM major > 2, search only these | |
| 1203 if (best_SM_arch > 2) | |
| 1204 { | |
| 1205 // If our device==dest_SM_arch, choose this, or else pass | |
| 1206 if (deviceProp.major == best_SM_arch) | |
| 1207 { | |
| 1208 max_compute_perf = compute_perf; | |
| 1209 max_perf_device = current_device; | |
| 1210 } | |
| 1211 } | |
| 1212 else | |
| 1213 { | |
| 1214 max_compute_perf = compute_perf; | |
| 1215 max_perf_device = current_device; | |
| 1216 } | |
| 1217 } | |
| 1218 } | |
| 1219 | |
| 1220 ++current_device; | |
| 1221 } | |
| 1222 | |
| 1223 return max_perf_device; | |
| 1224 } | |
| 1225 | |
| 1226 | |
| 1227 // Initialization code to find the best CUDA Device | |
| 1228 inline int findCudaDevice(int argc, const char **argv) | |
| 1229 { | |
|
304
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
1230 struct cudaDeviceProp deviceProp; |
| 291 | 1231 int devID = 0; |
| 1232 | |
| 1233 // If the command-line has a device number specified, use it | |
| 1234 if (checkCmdLineFlag(argc, argv, "device")) | |
| 1235 { | |
| 1236 devID = getCmdLineArgumentInt(argc, argv, "device="); | |
| 1237 | |
| 1238 if (devID < 0) | |
| 1239 { | |
| 1240 printf("Invalid command line parameter\n "); | |
| 1241 exit(EXIT_FAILURE); | |
| 1242 } | |
| 1243 else | |
| 1244 { | |
| 1245 devID = gpuDeviceInit(devID); | |
| 1246 | |
| 1247 if (devID < 0) | |
| 1248 { | |
| 1249 printf("exiting...\n"); | |
| 1250 exit(EXIT_FAILURE); | |
| 1251 } | |
| 1252 } | |
| 1253 } | |
| 1254 else | |
| 1255 { | |
| 1256 // Otherwise pick the device with highest Gflops/s | |
| 1257 devID = gpuGetMaxGflopsDeviceId(); | |
| 1258 checkCudaErrors(cudaSetDevice(devID)); | |
| 1259 checkCudaErrors(cudaGetDeviceProperties(&deviceProp, devID)); | |
| 1260 printf("GPU Device %d: \"%s\" with compute capability %d.%d\n\n", devID, deviceProp.name, deviceProp.major, deviceProp.minor); | |
| 1261 } | |
| 1262 | |
| 1263 return devID; | |
| 1264 } | |
| 1265 | |
| 1266 // General check for CUDA GPU SM Capabilities | |
| 1267 inline bool checkCudaCapabilities(int major_version, int minor_version) | |
| 1268 { | |
|
304
9755206813cb
helper_string.h for ANSI C
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
303
diff
changeset
|
1269 struct cudaDeviceProp deviceProp; |
| 291 | 1270 deviceProp.major = 0; |
| 1271 deviceProp.minor = 0; | |
| 1272 int dev; | |
| 1273 | |
| 1274 checkCudaErrors(cudaGetDevice(&dev)); | |
| 1275 checkCudaErrors(cudaGetDeviceProperties(&deviceProp, dev)); | |
| 1276 | |
| 1277 if ((deviceProp.major > major_version) || | |
| 1278 (deviceProp.major == major_version && deviceProp.minor >= minor_version)) | |
| 1279 { | |
| 1280 printf(" Device %d: <%16s >, Compute SM %d.%d detected\n", dev, deviceProp.name, deviceProp.major, deviceProp.minor); | |
| 1281 return true; | |
| 1282 } | |
| 1283 else | |
| 1284 { | |
| 1285 printf(" No GPU device was found that can support CUDA compute capability %d.%d.\n", major_version, minor_version); | |
| 1286 return false; | |
| 1287 } | |
| 1288 } | |
| 1289 #endif | |
| 1290 | |
| 1291 // end of CUDA Helper Functions | |
| 1292 | |
| 1293 | |
| 1294 #endif |