I’ve been using GPU accelerated computing for more than 10 years now. I write CUDA kernels in my sleep. I have seen 10s of frameworks and 10s of chip companies

I’ve been using GPU accelerated computing for more than 10 years now. I write CUDA kernels in my sleep. I have seen 10s of frameworks and 10s of chip companies come and go. I have NEVER even HEARD of Rocm, and neither has anyone in the GPU programming slack group I just asked. CUDA is absolutely a moat.