Offloading computation from CPU to GPU using Intel Visual Fortran Compiler with Intel Parallel Studio XE

[Gaby_S_]

Dear All,

I am currently running Fortran codes, parallelised with OpenMP using Intel Visual Fortran Compiler and Parallel Studio XE 2019 under Windows.

I am aware that Intel C++ Compiler provides a feature that enables offloading general purpose computations to processor graphics, referred to as Intel Graphics Technology. Is there a similar feature that enables computation offload to GPUs using Intel Visual Fortran embedded in Parallel Studio? I have come across an article https://software.intel.com/en-us/articles/the-heterogeneous-programming-model of 2012 in which this possibility is mentioned but haven't seen any more recent information on this topic.

Do you have any idea what would be Intel's hardware requirements to enable the offload to GPU? I recently heard that MIC architecture processors, such as Xeon Phi and Knights Landing  have been discontinued.

Your help will be much appreciated!

Regards,

Gaby S.

[Steve_Lionel]

I have successfully written and run Fortran programs that call the Intel OpenCL library to offload computations to a graphics card. See https://software.intel.com/en-us/forums/intel-visual-fortran-compiler-for-windows/topic/515753

[Gaby_S_]

Thank you very much for the link, Steve. This sounds very encouraging. I will look at the thread and try to do it as well.

Gaby