What I think the ‘make it or break it’ will be for folks is if we see NVENC, DLSS, CUDA support for NVK. The only way I see people who need Nvidia specific features ditching the proprietary drivers is if Nvidia releases proprietary blobs for them.
But as for me, I’m ditching the proprietary drivers as soon as NVK performs within 80% of the proprietary drivers.
Unfortunately I need CUDA. I hope either ZLUDA and/or actual CUDA works on NVK in the near future. Better yet AMD could release something that can compete with CUDA but that seems highly I. probable.
I hope either ZLUDA and/or actual CUDA works on NVK in the near future.
This is my hope as well. I do think that at least an attempt at CUDA support for NVK is planned, but if it is, it’s likely still a ways out. But who knows! They’ve been progressing so fast, it might come sooner than we think! (Assuming CUDA is even on the roadmap.)
Better yet AMD could release something that can compete with CUDA but that seems highly I. probable.
Unfortunately AMD’s focus doesn’t seem to be on Ray Tracing/AI or a CUDA alternative at the moment. But this would definitely be a welcomed feature.
What I think the ‘make it or break it’ will be for folks is if we see NVENC, DLSS, CUDA support for NVK. The only way I see people who need Nvidia specific features ditching the proprietary drivers is if Nvidia releases proprietary blobs for them. But as for me, I’m ditching the proprietary drivers as soon as NVK performs within 80% of the proprietary drivers.
NVK FTW!!
Unfortunately I need CUDA. I hope either ZLUDA and/or actual CUDA works on NVK in the near future. Better yet AMD could release something that can compete with CUDA but that seems highly I. probable.
This is my hope as well. I do think that at least an attempt at CUDA support for NVK is planned, but if it is, it’s likely still a ways out. But who knows! They’ve been progressing so fast, it might come sooner than we think! (Assuming CUDA is even on the roadmap.)
Unfortunately AMD’s focus doesn’t seem to be on Ray Tracing/AI or a CUDA alternative at the moment. But this would definitely be a welcomed feature.