i've been running inference on the 7900xtx using pytorch and rocm (installed directly from package managers, no manual fiddling) with great performance. no prob

i've been running inference on the 7900xtx using pytorch and rocm (installed directly from package managers, no manual fiddling) with great performance. no problem running the full flux1.dev model, for instance. haven't looked at training or fine-tuning yet.