Xavier Hallade
Xavier Hallade
Current CPU usage of OpenMM on GPU is quite high in Folding@Home: I do see 100% of one core on Intel GPUs with OpenCL and Nvidia GPUs with Cuda on...
call NDK_ABI=x86 ./configure_make_everything.sh to generate x86 binaries
On a system with several OpenCL platforms, if a platform has no device, one of the clGetDeviceIDs calls returns CL_DEVICE_NOT_FOUND. This currently prevents `getOpenCLDevices` from returning, throwing "OpenCL Error: Cannot...
Since https://github.com/intel/compute-runtime/commit/bbe11205e3605d9d4e53dfe9d1a2f8857f294d42, there are only suffixed versions of ocloc being built. Some distribution package specific code calls update-alternatives at install time to then symlink a version of ocloc: https://github.com/intel/compute-runtime/blob/b6038de550f3b7a411217ae6d4b4d8360d03d3a8/scripts/packaging/opencl/sles_15/SPECS/opencl.spec#L118 so...