I’m running two VMs on r5 1600 with one RX 580 per VM, with a GT 710 for Fedora Host. Looking glass itself is running without a hitch, but when both VMs are running with two instances of looking glass on host, the framerate drops on both instances.
I’m not too sure if this is a bottleneck of my GT 710 running on pciex1 (or the card itself being too slow for both), a sata bottleneck between two drives, or a CPU bottleneck.
My question is, to someone a little more knowledgeable than me, how can I keep looking glass’ framerate up for both instances?
The frame rate (FPS) or UPS?
The FPS dips to 40 or 30 when both are running at 960x1080 resolution. The size of the memory file is 32 mb I believe for 1920x1089
Again, please confirm if this is the FPS or UPS dipping? If you are unsure what this is, please be sure to turn on the FPS display in the LG client.
With a single VM running at 960x1080 at a memory size of 32, it runs at ~96 FPS with a UPS of 0. A single VM runnin at 960x1080 at a memory size of 16, it runs at ~94 FPS with a UPS of 0. Both VMs run at a memory size of 16 runs at ~50FPS with a UPS of 0. I noticed that when I copied a file from one VM to another, the FPS dropped to ~30 FPS on both until the file transfer was finished
Ok, the performance issue then is not the VM but rather the host GPU (GT 710), UPS is how many frames per second are being sent from the VM, if it’s reading 0 then there is nothing being sent and the client application is just showing a static image (cursor is rendered seperate)
FPS drop during the file copy is simply because you are now also getting new frame updates from the VM, you would see the same behaviour if you ran any program which was continually updating the screen (ie, a game or movie).
Please be sure you have the official Nvidia drivers installed and are not using nouveau. The issue is with your GT 710’s performance for this task.
So if the GT 710 simply doesn’t have the horsepower, do you think that a GT 1030 will?
I am sorry I wouldn’t know, you would need to test. Most people do not run multiple LG clients.
Do you know if Looking Glass’ performance is directly tied to Host GPU performance? or are there more prominent factors?
Usually it’s bound by the Windows Capture API performance, and guest GPU performance.
Thanks for all the help.
I download the official nVidia driver and performanced dramatically increased, but not to a full 60fps. Usage of the gt 710 was around 95, so I’m assuming that that card is the bottleneck here.
Quite possible