I can't find a clear answer about level of support for GPU compute offload on Linux. It seems that the newest 6th generation CPUs are not supported both on Windows and Linux. Support ends with 5th generation CPUs as stated in Release Notes for Intel Parallel Studio XE 2017. Is that accurate? On the other hand, OpenCL support is better under Windows, where it reaches OpenCL 2.0 on 6th generation CPUs, whereas Linux stops at OpenCL 1.2 on 5th generation CPUs. Will there be a parity in OpenCL support anytime soon?
↧