I am a FORTRAN programmer, with Parallel studio XE 2019 (recently updated to 2020) installed in my pc. I'm interested in the oneAPI interface for extending my code to Nvidia GPU hardware.
What is the suggested configuration, maintaining my previous FORTRAN code? Is it possible to mix oneAPI libraries (MKL, OpenCL) with Fortran?
The tools to access Intel accelerators are only available using the oneAPI beta Toolkits. This link has information on downloading the appropriate toolkits. With these kits MKL and other libraries are available.
Neither PSXE 2019 nor PSXE 2020 support offloading to accelerators.
Only Intel accelerators are supported.
Thank you very much for your help.
I have downloaded the OneAPI base toolkit for installing, by the way the installer alert me that I will loose my previous VS integration of the PSXE. I'm confused, should I go on with that? Is the new compiler functionality the same of the PSXE? What happens to the other tools (like debugger, VTune, etc.)? If I choose to not integrate it with VS, I still have console access to the two different Fortran Compilers?
I still can't understand what is the best configuration...
The OneAPI toolkit VS integration will also have integration of tools like Vtune, Advisor, Intel Inspector, debugger, etc.
And if you choose not to integrate, then you can still access them using the console, the only thing is that you have to source the environment at every instance before using the tools from the OneAPI base kit. After sourcing only you can able to use compilers and other tools from the OneAPI toolkit.