matrix_cuda.cu 2.0 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273
  1. /* StarPU --- Runtime system for heterogeneous multicore architectures.
  2. *
  3. * Copyright (C) 2011 INRIA
  4. *
  5. * StarPU is free software; you can redistribute it and/or modify
  6. * it under the terms of the GNU Lesser General Public License as published by
  7. * the Free Software Foundation; either version 2.1 of the License, or (at
  8. * your option) any later version.
  9. *
  10. * StarPU is distributed in the hope that it will be useful, but
  11. * WITHOUT ANY WARRANTY; without even the implied warranty of
  12. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
  13. *
  14. * See the GNU Lesser General Public License in COPYING.LGPL for more details.
  15. */
  16. #include <starpu.h>
  17. #include "../test_interfaces.h"
  18. extern struct test_config matrix_config;
  19. __global__ void matrix_cuda(int *val, unsigned n, int *err, int factor)
  20. {
  21. unsigned i = blockIdx.x*blockDim.x + threadIdx.x;
  22. if (i >= n)
  23. return;
  24. if (val[i] != i*factor)
  25. *err = 1;
  26. else
  27. val[i] = -val[i];
  28. }
  29. extern "C" void test_matrix_cuda_func(void *buffers[], void *args)
  30. {
  31. int factor;
  32. int *ret;
  33. int *val;
  34. cudaError_t error;
  35. unsigned int nx, ny, n;
  36. nx = STARPU_MATRIX_GET_NX(buffers[0]);
  37. ny = STARPU_MATRIX_GET_NY(buffers[0]);
  38. n = nx * ny;
  39. unsigned threads_per_block = 64;
  40. unsigned nblocks = (n + threads_per_block-1) / threads_per_block;
  41. factor = *(int *) args;
  42. val = (int *) STARPU_MATRIX_GET_PTR(buffers[0]);
  43. error = cudaMalloc(&ret, sizeof(int));
  44. if (error != cudaSuccess)
  45. STARPU_CUDA_REPORT_ERROR(error);
  46. error = cudaMemcpy(ret,
  47. &matrix_config.copy_failed,
  48. sizeof(int),
  49. cudaMemcpyHostToDevice);
  50. if (error != cudaSuccess)
  51. STARPU_CUDA_REPORT_ERROR(error);
  52. matrix_cuda<<<nblocks,threads_per_block,2,starpu_cuda_get_local_stream()>>>(val, n, ret, factor);
  53. error = cudaMemcpy(&matrix_config.copy_failed,
  54. ret,
  55. sizeof(int),
  56. cudaMemcpyDeviceToHost);
  57. if (error != cudaSuccess)
  58. STARPU_CUDA_REPORT_ERROR(error);
  59. cudaFree(ret);
  60. cudaStreamSynchronize(starpu_cuda_get_local_stream());
  61. }