If I’m using a dGPU instead of an iGPU for the linux side of looking glass, how much VRAM is needed for it to run 4K@60Hz without issues? What about for 4K@144Hz? I’m thinking of getting an older, used dGPU for cheap to avoid this iGPU lack of VRAM problem but want to make sure whatever I get has enough specs to make 4K gaming performant.
In future please do not necro or hijack old posts.
how much VRAM is needed for it to run 4K@60Hz without issues
It has nothing to do with ram, but rather bandwidth. iGPUs are not suited at all for this kind of usage as they share system RAM which effectively halves the available RAM bandwidth compared to a system with two dGPUs.
This is the best way to go, any GPU made in the last 10 years that supports a resolution of 4K @ 144Hz will do, the GPU is only doing very basic work (drawing the already rendered frame) and doesn’t need to be anything special.
I am using an RX550 with 2GB RAM for this purpose. It works perfectly fine for the reasons gnif explained. I am going to upgrade to a RX7600, but only because it supports AV1 video decoding. For it only to display the LG window it does not need much performance.
If I may just share a success story: I am right now using looking glass quite smoothly at 4K120 using my 7950X’s iGPU on the Host and an nvidia 3060 on the Guest. Have not tried gaming yet, but the desktop is very smooth. Not quite as smooth as native, but better than native 60Hz 
I’m on fedora 39, compiled client from git master, and using the latest Windows host binary.
1 Like
Unless you’re a developer you should not be using git for LG. You can obtain the latest builds from the LG website including a pre-prepared source archive for building the client.
We do not support people using git because they often do not understand how to properly update sub-modules when we push changes causing invalid/erroneous bug reports.
1 Like