Without question, nvidia. The sad reality here is that the best performance for ML on a 5700xt is to boot into windows and use the beta for directml under WSL2

Without question, nvidia. The sad reality here is that the best performance for ML on a 5700xt is to boot into windows and use the beta for directml under WSL2 with TF1. It pains me to say that as someone who has been a linux purist for over a decade. We’re literally just waiting for AMD to release blobs, and we’ve been waiting over a year.