GCC's Ada Frontend Seeing More Work On NVIDIA CUDA Support
In the past born out of academia there's been CUDA Ada bindings. There has also been Ada/SPARK GPU programming initiatives in the past with various APIs. This latest still ongoing effort is wiring up the GCC Ada front-end with CUDA support.
Going back to this summer there have been stubs for CUDA pragmas in the Ada code while this week more code is being merged on this front.
Adacore developers have been making various commits and like implementing expansion of CUDA_Execute and initialization of the CUDA runtime.
The Ada front-end is using -gnatd_c as the flag for targeting CUDA. Like with other GCC front-ends supporting CUDA, ultimately NVPTX will be dumped that can then be consumed by the proprietary NVIDIA graphics/compute stack. However, at this point it appears all the infrastructure work is still being taken care of and isn't yet viable for programmers wanting to begin writing Ada code for NVIDIA GPU execution with CUDA. It will be interesting to see this GCC Ada + CUDA support come together though considering GCC 11 feature development ends in one month, it's highly unlikely this will be in suitable shape for this next GCC release.