In an effort to get a setup going where I can virtualize windows with ~native performance, so I can fully stay on linux for my actual OS, I've been looking into vGPU tools. Since I know (and have tested) that Hyper-V on Win10 can virtually share a GPU between host and guest, I assumed it must be doable on linux, and while that does seem to be the case, I'm having a lot of trouble making sense of the current state of things.
The biggest thing for me is GVM, which is a technology mentioned and linked in the setup guide... leading to a dead link at linux-gvm.org. Does anyone know where documentation is currently hosted for that? I find it difficult to dig through related repos that link back to the same site, and all the Kali tool links for the thing they call GVM.
If there's anyone around here more involved (or more knowledgeable in general) than me (not hard, probably most of y'all), maybe look into it if vGPU stuff interests you! Then come back here and let me know what I was missing cause I feel stupid trying to understand mdev and GVM vGPU differences.
That aside, based on open-iov's own guide having warnings to follow the other one, and said other one having some VERY unclear steps (Section 5.3.4 referring to previously made yaml edits when none were mentioned), I imagine that tool is not too actively maintained. Does anyone know where I might find some outline or guide to setting that system up from scratch?
As it turns out, GVM is still a live project! Not sure why their site is down, seems like just a lot of old documentation links to a site they decided not to use. After asking around the discord it seems like LibVF.IO isn't a dead project exactly, but the only way to work with up-to-date things is to manually find all the tools it uses and set everything up yourself. The teams working on all of that stuff are a lot more focused on functionality than public guides and documentation.
tldr; The space isn't dead! Just low-population, and busy making things actually work.
I for one am pretty excited for this stuff to become more accessible, anyone else? I'm trying to wrap my head around the whole setup they envision so I can maybe help contribute, so if anyone is in the space and wants to give a run down on the whole process of getting a vGPU setup and into a VM, that would be a good addition to this thread!