vector_scal_opencl.c 2.2 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768
  1. /* StarPU --- Runtime system for heterogeneous multicore architectures.
  2. *
  3. * Copyright (C) 2010, 2012, 2013, 2016 CNRS
  4. * Copyright (C) 2010 INRIA
  5. * Copyright (C) 2011, 2014 Université de Bordeaux
  6. *
  7. * StarPU is free software; you can redistribute it and/or modify
  8. * it under the terms of the GNU Lesser General Public License as published by
  9. * the Free Software Foundation; either version 2.1 of the License, or (at
  10. * your option) any later version.
  11. *
  12. * StarPU is distributed in the hope that it will be useful, but
  13. * WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
  15. *
  16. * See the GNU Lesser General Public License in COPYING.LGPL for more details.
  17. */
  18. /*
  19. * This example complements vector_scal.c: here we implement a OpenCL version.
  20. */
  21. #include <starpu.h>
  22. extern struct starpu_opencl_program opencl_program;
  23. void scal_opencl_func(void *buffers[], void *_args)
  24. {
  25. float *factor = _args;
  26. int id, devid;
  27. cl_int err;
  28. cl_kernel kernel;
  29. cl_command_queue queue;
  30. /* length of the vector */
  31. unsigned int n = STARPU_VECTOR_GET_NX(buffers[0]);
  32. /* OpenCL copy of the vector pointer */
  33. cl_mem val = (cl_mem)STARPU_VECTOR_GET_DEV_HANDLE(buffers[0]);
  34. id = starpu_worker_get_id_check();
  35. devid = starpu_worker_get_devid(id);
  36. err = starpu_opencl_load_kernel(&kernel, &queue, &opencl_program, "vector_mult_opencl", devid);
  37. if (err != CL_SUCCESS) STARPU_OPENCL_REPORT_ERROR(err);
  38. err = clSetKernelArg(kernel, 0, sizeof(n), &n);
  39. err |= clSetKernelArg(kernel, 1, sizeof(val), &val);
  40. err |= clSetKernelArg(kernel, 2, sizeof(*factor), factor);
  41. if (err) STARPU_OPENCL_REPORT_ERROR(err);
  42. {
  43. size_t global=n;
  44. size_t local;
  45. size_t s;
  46. cl_device_id device;
  47. starpu_opencl_get_device(devid, &device);
  48. err = clGetKernelWorkGroupInfo (kernel, device, CL_KERNEL_WORK_GROUP_SIZE, sizeof(local), &local, &s);
  49. if (err != CL_SUCCESS) STARPU_OPENCL_REPORT_ERROR(err);
  50. if (local > global) local=global;
  51. err = clEnqueueNDRangeKernel(queue, kernel, 1, NULL, &global, &local, 0, NULL, NULL);
  52. if (err != CL_SUCCESS) STARPU_OPENCL_REPORT_ERROR(err);
  53. }
  54. starpu_opencl_release_kernel(kernel);
  55. }