I’m interested in the current status and plans on OpenACC and OpenMP offloading support in Flang. Specifically, I’m interested in GPU reductions in both OpenACC and OpenMP, and independent clause in OpenACC. Is there anyone actively involved in the development of these features or plan to do so in future?
Thanks for your interest in OpenACC/OpenMP Flang work.
OpenACC work is currently on hold due to a shift in focus for finishing the Flang work.
We are making progress with the OpenMP offloading work. A few engineers from AMD are working on this at the moment. We are also generally working on the reduction clause. This is currently limited to a few types and scalar context. We have not done anything special for GPUs.
For OpenMP, we have biweekly meetings to share and update status. https://docs.google.com/document/d/1yA-MeJf6RYY-ZXpdol0t7YoDoqtwAyBhFLr5thu5pFI/edit
The Project Management sheet is OMP builder + Flang OMP tasks - Google Sheets
You are welcome to attend and contribute.
@kiranchandramohan thanks for sharing this information!