Some tougher suggestions from Red Hat's @airlied for managing open-source community projects. "Discourage non-engaged companies/singularly product focused companies"

Here are @airlied's proposals for the UXL Foundation. I think they're great suggestions

We're also making huge progress in AI, with major AI frameworks like TensorFlow and PyTorch now supporting oneAPI and SYCL. I realise a lot of people claim to support these frameworks via AI graph compilers, but these graph compilers give great performance but at the cost of significantly reduced capabilities, whereas we're getting upstream support integrated into the frameworks that is highly feature-rich and supported as well as high performance

We've been able to get a lot of adoption in scientific software, enabling high performance scientific applications work on multiple processor architectures now and in the future.

We've built up a great set of organizations of members who are contributing, porting to their hardware, porting their software to oneAPI, or driving future improvements. You can join to, even with free membership options. If you look carefully, you'll see we're enabling software acceleration on some really diverse processor architectures, while also supporting some very important software