5 ms·
Step 1: Run on NVIDIA gpus until it works just as well as real CUDA. Step 2: Port to other GPUs. At least I assume that is the plan.
by jedberg 2y ago
Step 1: Run on NVIDIA gpus until it works just as well as real CUDA.
Step 2: Port to other GPUs.
At least I assume that is the plan.
- chii 2y ago> Step 2: Port to other GPUs. why not do this first? because the existing closed sourced CUDA already runs well on nvidia chips. Replicating it with an open stack, while ideologically useful, is going to sap resources away from the porting of it to other GPUs (where the real value can be had - by stopping the nvidia monopoly on ai chips).
- jfim 2y agoI'm not involved with the project but I'd assume it's helpful as a reference implementation. If there's a bug on a non-nvidia GPU, the same test case can be run on Nvidia GPUs to see if the bug is in common code or specific to that GPU vendor.