| | 1 | = OpenCL Applications = |
| | 2 | |
| | 3 | You can run OpenCL applications under BOINC. |
| | 4 | |
| | 5 | == Application requirements == |
| | 6 | |
| | 7 | Your application must call |
| | 8 | {{{ |
| | 9 | int boinc_get_opencl_ids(cl_device_id*, cl_platform_id*); |
| | 10 | }}} |
| | 11 | |
| | 12 | This returns the OpenCL platform and device IDs for the GPU |
| | 13 | that your app should use. |
| | 14 | |
| | 15 | == Scheduler requirements == |
| | 16 | |
| | 17 | The following plan classes are defined in '''sched/sched_customize.cpp''': |
| | 18 | |
| | 19 | '''opencl_nvidia_101''':: |
| | 20 | For app versions that run on NVIDIA GPUs using OpenCL 1.1, using at most 256MB of GPU RAM. |
| | 21 | '''opencl_ati_101''':: |
| | 22 | For app versions that run on ATI GPUs using OpenCL 1.1, using at most 256MB of GPU RAM. |
| | 23 | |
| | 24 | You can modify sched_customize.cpp to change these parameters or add |
| | 25 | your own plan classes. |