Ekansh Jain
Ekansh Jain
Is this related to issue #588 because the problem seems similar?
I think an approach similar to the one in this [OpenCL Wrapper](https://github.com/ProjectPhysX/OpenCL-Wrapper/blob/master/src/opencl.hpp) could be used. Exact info: `Nvidia Pascal and newer GPUs with driver>=520.00 don't report cl_khr_fp16, but do support...
Here are the relevant lines of code for this approach, the ones with the arrow are the ones to look at: ``` —> if(is_gpu) nvidia_compute_capability = 10u*(uint)cl_device.getInfo()+(uint)cl_device.getInfo(); const bool nvidia__32_cores_per_cu...
Is there an error code returned from `clGetEventProfilingInfo`, for example profiling is not enabled? Also, the device has a specific precision for its events and can be queried via: `CL_DEVICE_PROFILING_TIMER_RESOLUTION`,...
What if pointers to functions were added as public member variables of the context class. Then, the user will supply some SVM selector function which based on the context will...
If you parse for types with the category `define`, then get the `` tag, and then get the value after the closing name tag, you can parse from the xml....
Hi, can you please let me know if you are willing to remove CUDA support so that I can work towards doing that if I have the time.
@CNugteren do you want to continue on this removal or should we wait a few more months?
I was working on PR #649 and found that CMake can't detect OpenCL 3.0 before version 3.24, should we put this off then?
Created pull request #583 to report that 'ldc=0' is not supported with a new error code.