Can you try to make the window size smaller? Like much smaller. Maybe tor browser has limited size of your huge 4k screen?
still having issues. Should be an aspect ratio concern because my laptop screen is native 1080p, 16:9
Iām months late to the thread, but Iād like to add that I completely disagree with unman and RIP.
Being forced to watch videos on a separate device not running Qubes OS can be a security and OPSEC risk. I know that people will ridicule this notion with āif itās that important download and watch offlineā, but the point is, if one has issues watching sensitive media/content on Qubes OS, then the user simply has no choice but to reduce their security by not using Qubes OS, which might affect their operational security.
To note: this is not the fault of Qubes dev team. They have done a fantastic job, and as DemiMarie has mentioned across forums and mailing lists alike, existing implementations of passing virt-io powered GPU devices as PCIe to VMs is not secure. The latest mechanisms are VirGL and Venus, which work to provide a virtual GPU for OpenGL/Vulkan calls respectively, and they work primarily in the user space with little to no security boundary, making them completely unacceptable for use in Qubes OS.
Native DRM contexts are in development at both Intel and AMD, and when that is done, it will be integrated with virt-host (QEMU is already working on it, and itās used by crosvm too, with an acceptable security footprint). This alongside SR-IOV will be the recommended GPU acceleration mechanisms for guests going forward.
The most unfortunate part of this is that there is a major dependency on Wayland, and as we know Qubes OS doesnāt use Wayland. I believe this is a MASSIVE security flaw, even if itās just dom0 using it. But I will not criticize the invisible things lab for I believe that they are doing what they can.
Overall, I expect a 2-5 year wait time to get GPU acceleration supported on Qubes OS. This will adversely affect some people, but it is a technical limitation of Qubes OS. There is nothing that can be done at the moment without support from Intel and AMD and Qubes moving to Wayland (work has already started on this front on Github: GPU acceleration Ā· GitHub
I will try ytfzf
and see if playback is any better.