On Saturday, 20 January 2018 09:40:36 UTC, Foppe de Haan wrote: > On Saturday, January 20, 2018 at 9:38:06 AM UTC+1, Alex Dubois wrote: > > On Thursday, 18 January 2018 22:56:10 UTC, Tom Zander wrote: > > > On Sunday, 14 January 2018 08:12:24 CET r...@tuta.io wrote: > > > > Is qubes able to use the computing power of the gpu or is the type of > > > > gpu > > > > installed a waste in this issue? > > > > > > Relevant here is an email I wrote recently; > > > https://groups.google.com/forum/#!msg/qubes-devel/40ImS390sAw/Z7M0E8RiAQAJ > > > > I'll reply in that thread about this to stay in topic. > > > > But in few words: Not possible until GPU virtualization to have a trustable > > solution. > > > > > > > > The context is a GSoC proposal proposal to modernize the painting > > > pipeline of Qubes. > > > > > > Today GL using software uses [llvmpipe] to compile and render GL inside > > > of > > > a Qube, completely in software and then push the 2d image to dom0. > > > This indeed wastes the GPU. > > > > > > > > > [llvmpipe]: > > > https://groups.google.com/forum/#!msg/qubes-devel/40ImS390sAw/Z7M0E8RiAQAJ > > > > > > -- > > > Tom Zander > > > Blog: https://zander.github.io > > > Vlog: https://vimeo.com/channels/tomscryptochannel > > Since I am unable to estimate the security aspects of any given approach, and > you do, have you seen this approach? > https://forum.level1techs.com/t/looking-glass-guides-help-and-support/122387
I am not member of the Qubes core team, I am an avid user/developer and believer :) so my view is only mine... The project you mention is doing a great job (for a VMWare workstation type set-up), however as far as I understood the copy is from/to the same GPU. This is where I am NOT comfortable with. As explained the client VM would issue processing requests to the GPU (and may abuse it). However, using their work to copy from one GPU (assigned to ONE VM) to Dom0 GPU could be good. However you still have the problem with the BW on the bus (luckily depending on your hardware build 2 different buses (your 2 cards are on different PCIe lines). You will not get 144Hz but 60Hz is within reach. Temptation to compress the stream will be there, the decompression code will be in the attack surface. -- You received this message because you are subscribed to the Google Groups "qubes-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to qubes-users+unsubscr...@googlegroups.com. To post to this group, send email to qubes-users@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/qubes-users/553b21c0-cc0c-47f6-b9e1-2ba03762164f%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.