> Q: How far should Debian go along the way for supporting hardware > acceleration solutions like CUDA?
> Choice 1: this game belongs to the big companies. we should offload > such burden to third-party providers such as Anaconda. > Choice 2: we may try to provide what the users need. > Choice 3: <write down yours> > --- I am alos for <2> We badly need rocm in order to have a viable alternative to NVIDIA in Debian. To my opinion providing all these stack will contribute a lot to make debian-science relevant for the next couple of years. More and more softwares in my area (xray data treatment) are using pytorch or tensorflow. If we can not install this via Debian, it will be difficult for us to request budget in order to improve Debian via subcontracting. I agree also with you that a pre-requisite is to provide these hardware to the community in order to do developpement via porterbox, autopkgtest, salsa, build Do we know if the salsa cloud provider propose runners with GPU's ? cheers Frederic
