fblock_opencl.c 2.6 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970
  1. /* StarPU --- Runtime system for heterogeneous multicore architectures.
  2. *
  3. * Copyright (C) 2010, 2011, 2012, 2013 Centre National de la Recherche Scientifique
  4. * Copyright (C) 2011, 2014 Université de Bordeaux 1
  5. *
  6. * StarPU is free software; you can redistribute it and/or modify
  7. * it under the terms of the GNU Lesser General Public License as published by
  8. * the Free Software Foundation; either version 2.1 of the License, or (at
  9. * your option) any later version.
  10. *
  11. * StarPU is distributed in the hope that it will be useful, but
  12. * WITHOUT ANY WARRANTY; without even the implied warranty of
  13. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
  14. *
  15. * See the GNU Lesser General Public License in COPYING.LGPL for more details.
  16. */
  17. #include <starpu.h>
  18. #define CHECK_CL_SET_KERNEL_ARG(kernel, n, size, ptr) \
  19. do \
  20. { \
  21. int check_err; \
  22. check_err = clSetKernelArg(kernel, n, size, ptr); \
  23. if (check_err != CL_SUCCESS) \
  24. STARPU_OPENCL_REPORT_ERROR(check_err); \
  25. } while (0)
  26. extern struct starpu_opencl_program opencl_program;
  27. void opencl_func(void *buffers[], void *cl_arg)
  28. {
  29. int id, devid, err;
  30. cl_kernel kernel;
  31. cl_command_queue queue;
  32. cl_event event;
  33. int *factor = cl_arg;
  34. cl_mem block = (cl_mem)STARPU_BLOCK_GET_DEV_HANDLE(buffers[0]);
  35. unsigned offset = STARPU_BLOCK_GET_OFFSET(buffers[0]);
  36. int nx = (int)STARPU_BLOCK_GET_NX(buffers[0]);
  37. int ny = (int)STARPU_BLOCK_GET_NY(buffers[0]);
  38. int nz = (int)STARPU_BLOCK_GET_NZ(buffers[0]);
  39. unsigned ldy = STARPU_BLOCK_GET_LDY(buffers[0]);
  40. unsigned ldz = STARPU_BLOCK_GET_LDZ(buffers[0]);
  41. id = starpu_worker_get_id();
  42. devid = starpu_worker_get_devid(id);
  43. err = starpu_opencl_load_kernel(&kernel, &queue, &opencl_program, "fblock_opencl", devid);
  44. if (err != CL_SUCCESS) STARPU_OPENCL_REPORT_ERROR(err);
  45. CHECK_CL_SET_KERNEL_ARG(kernel, 0, sizeof(block), &block);
  46. CHECK_CL_SET_KERNEL_ARG(kernel, 1, sizeof(offset), &offset);
  47. CHECK_CL_SET_KERNEL_ARG(kernel, 2, sizeof(nx), &nx);
  48. CHECK_CL_SET_KERNEL_ARG(kernel, 3, sizeof(ny), &ny);
  49. CHECK_CL_SET_KERNEL_ARG(kernel, 4, sizeof(nz), &nz);
  50. CHECK_CL_SET_KERNEL_ARG(kernel, 5, sizeof(ldy), &ldy);
  51. CHECK_CL_SET_KERNEL_ARG(kernel, 6, sizeof(ldz), &ldz);
  52. CHECK_CL_SET_KERNEL_ARG(kernel, 7, sizeof(*factor), factor);
  53. {
  54. size_t global=nx*ny*nz;
  55. err = clEnqueueNDRangeKernel(queue, kernel, 1, NULL, &global, NULL, 0, NULL, NULL);
  56. if (err != CL_SUCCESS) STARPU_OPENCL_REPORT_ERROR(err);
  57. }
  58. starpu_opencl_release_kernel(kernel);
  59. }