I have two graphic cards on my machine. A P2000 and a 2080 ti. In the pref for both CUDA and Optix, I choose the 2080 because it has 11 gigs of vram. But Blender doesn’t use it at all. I used the watch -n0.1 nvidia-smi command to monitor the cards and it’s clear that the 2080 is inactive. I know the card is OK because it’s working with Substance.
Any idea?
Hi Bob !
Do you use it for rendering with Cycles ? with Eevee/Viewport it uses the display graphic card…
I use a 2080 ti and a 2070 super and can get cycles to use both but I also use windows and stick to the studio drivers from nvidia. On windows there is an option to choose which gpu is used to render open gl on (viewport). I have looked into it and for the most part (minus the hardware differences) blender doesnt care about the quadro drivers. in auto CAD and other Autodesk software it does make a difference.
I use only Cycles. If I look at the stats at the bottom of the screen, I can see that the VRAM is only 5GIG. It doesn’t see the 2080 at all.
So what is happening is that since the p2000 only has 5 gigs that is what blender is using. In older versions of blender the lowest vram amount is what is displayed if using 2 cards. You may need to choose which gpu is used for open gl and which is only used for rendering. There are also some driver/package issues with linux at times. There can also be an issue when using blender with a gpu that is not connected to a display.
https://docs.blender.org/manual/en/latest/troubleshooting/gpu/linux/other.html
Apparently it’s a problem with Teradici, le software we use for remote work.
Hmmm…well there you go i guess. Was the software using the 2080?
Nop! Not even! It just blocks the 2080. I guess it’s jealous.