topic Re:Resulting executable and CPU/GPUs in Intel® oneAPI DPC++/C++ Compiler

Resulting executable and CPU/GPUs

eldiener — Fri, 25 Dec 2020 02:23:18 GMT

If I create a resulting executable with the Data Parallel C++ compiler is that executable limited in any way when being run on non-Intel CPUs or GPUs ?

Re: Resulting executable and CPU/GPUs

RahulV_intel — Mon, 28 Dec 2020 11:33:44 GMT

Hi,

DPC++ compilation can be performed in two ways:

JIT (Just in Time) compilation (Online compilation)
AOT (Ahead of Time) compilation (Offline compilation)

JIT compilation loads a generic SPIRV image into the executable. The machine code will be generated "on the fly" for the device that is being targeted.

AOT compilation loads a device-specific image into the executable. The machine code will be generated during the compilation phase ("Ahead of time") for the target device.

In short, the JIT-compiled executables should be portable (provided you have installed the same oneAPI version across both machines).

More information here:

https://software.intel.com/content/www/us/en/develop/documentation/oneapi-programming-guide/top/programming-interface/cpu-flow.html

Currently, oneAPI doesn't support non-Intel accelerators. You can check the hardware requirements here:

https://software.intel.com/content/www/us/en/develop/articles/intel-oneapi-base-toolkit-system-requirements.html

However, the open Github release of DPC++ has limited support for non-Intel accelerators. https://github.com/intel/llvm/blob/sycl/sycl/doc/GetStartedGuide.md

Thanks,

Rahul

Re:Resulting executable and CPU/GPUs

RahulV_intel — Tue, 05 Jan 2021 06:59:48 GMT

Hi,

Any updates on this? If the solution provided helped, could you please let me know if I can close this thread from my end?

Thanks,

Rahul

Re:Resulting executable and CPU/GPUs

RahulV_intel — Tue, 12 Jan 2021 09:36:45 GMT

Hi,

I have not heard back from you, so I will go ahead and close this thread from my end. Feel free to post a new query if your issue persists.

Regards,

Rahul