I’m running OpenCL on a Tesla M2050 in a machine with CUDA 5.0.35 installed. I’m seeing huge overhead from calling clCreateContext (sometimes > 1 second). Has anyone seen this before, or have any idea what might be causing it? I expect some initialization overhead but 1 second seems excessive.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Context creation on C2050 is 1-2 seconds? Why so long? | 4 | 4844 | August 11, 2010 | |
| clCreateContext with Fermi | 3 | 3609 | October 6, 2010 | |
| Is there a maximum number of contexts per GPU encoded into the driver? | 4 | 3396 | February 17, 2018 | |
| clCreateContext fails with error code -6 | 0 | 8149 | June 7, 2011 | |
| CUDA setup times (create context, malloc, destroy context) some measurements included | 19 | 23396 | July 8, 2011 | |
| Speed of creating a context | 3 | 957 | March 21, 2014 | |
| clCreateKernel has expensive overhead? | 3 | 8398 | January 5, 2010 | |
| Newbie help, OpenCL, creating context, nothing happens | 0 | 60 | January 29, 2025 | |
| CL_DEVICE_NOT_AVAILABLE after 12 calls clCreateContextFromType fails | 6 | 3089 | July 6, 2011 | |
| 260.19.06 32-bit Linux driver: CL_OUT_OF_RESOURCES on clCreateContext on code which ran fine with 19 | 0 | 11004 | November 3, 2011 |