Adding information on how to use oneDPL on other vendor platforms

[timmiesmith]

The PR adds notes to the Getting Started Guide on installing the oneAPI for NVIDIA GPUs plugin as well as a reference to the System Requirements section of the Developer's guide for the installation requirements to run oneDPL on CPU. This is meant to provide the documentation requested in uxlfoundation/open-source-working-group#51 on how to use oneDPL on different vendor hardware platforms.

[mmichel11]

The uxlfoundation issue this is addressing requests, "... and also how to build from source for different processor targets".

Should we add a link to https://intel.github.io/llvm/GetStartedGuide.html#build-dpc-toolchain which contains instructions on how to build the CUDA / HIP DPC++ toolchain from source?

[timmiesmith]

The uxlfoundation issue this is addressing requests, "... and also how to build from source for different processor targets".

Should we add a link to https://intel.github.io/llvm/GetStartedGuide.html#build-dpc-toolchain which contains instructions on how to build the CUDA / HIP DPC++ toolchain from source?

That I don't think we need to include. oneDPL with the OpenMP and/or oneTBB backends compile with multiple compilers so the DPC++ compiler isn't necessarily required, and I think since we've got information on CMake and the oneTBB and OpenMP requirements in the file we have the information needed to cover processor targets.

[mmichel11]

LGTM