Return to

Tesla V100 Gaming Possibility?

Hi all.
I want to preface this by saying I’m not particularly knowledgeable about VFIO or IOMMU so please point out anything I seem to be misunderstanding.

For context, I’m looking to get a Volta GPU for my workstation/main pc to do CUDA accelerated compute projects using double precision floats (orbital stuff) probably in popOS, maybe just Ubuntu. I also intend to do some gaming on it as well in my downtime. I’ve been looking for a Titan V or Quadro GV100 primarily for this build, but realized a Tesla V100 would work well for this as well.

Now to my question :slight_smile:
I’ve looked around (to no real avail) at how to possibly game on said Tesla V100, possibly using Looking Glass or something similar in conjunction with the iGPU on a Ryzen APU. Does anyone have experience making this work (either using a VM and Looking Glass or possibly natively in Linux)?

Thank you in advance!

A tesla is a gpu yes, but it’s not exactly for 3d objects…

On windows at least it works with the k40 on lastest drivers

with a little encouragement from the registry

Idk about linux

1 Like

I have the same interest. can you go into more detail. I have background in medicine but learning computer. want to build 3d models for healthcare.

currently making a thread to pair with a video with detailed instructions, I’ll put you on the tag list when it’s finished
but first things first if you’re using just one you’ll mostly be fine with a modern ryzen/thread ripper system but if you’re using more than one you’ll want a epyc simply because consumer bar space just can’t handle multiple tesla’s


Thanks! I’d appreciate it. My plan is to have 1 so I figure even an 8 core is fine for it. My main thought as of late is whether something like Looking Glass would be hindered by the increased memory latency on Ryzen for APUs compared to stock Intel iGPUs.

It seems we’re in the same boat–probably advisable to check the thread mentioned above.

can’t really help you with linux and looking glass, I’m a pleb so I can only do windows at the moment
The hardest part is getting hardware that supports the necessary bar space and has options like above 4g decoding

for windows what it amounts to is enableing Optimus on hardware its not meant to, waiting for the video to go live before I post the guide

I’ll also cover how to overclock on the tesla’s as well, you can gennerally get 200-275mhz out of them with stock power limit and voltage


Regardless I’m sure it’ll be valuable to many people :slight_smile:

1 Like

to do this you will need 2 gpu’s apparently the passthrough doesn’t work if you try to use the igpu as the main display adapter.
reason being the igpu isnt connected to the pci-e. and passthrough is a pci-e based feature.

so you will need another card.
just to run as a display adapter.
as for how powerful it needs to be?.
no idea.
but i would think it would need a gig of display ram at least to keep up with modern displays.
other than that, you might be able to get away with a really cheap video card rather than a full blow gaming card.

some on who has run a config like this might be able to help there?

There seems to be people who have made it work:

Not to very good results, but it works, and surely can be improved upon.

I currently run ITX so I would need to bifurcate the PCIe slot to make that work, or do an m.2 to PCIe adapter. But the IOMMU groupings on my board put all the bifurcated slots into the same group so…

yes mate it sorta works if. but sorta works isnt working properly and ready for lemans so to speak.
you will literally spend more time diagnosing bugs than playing games on it.
as it really is a work in progress and unless your absolutely sure you want this burden :wink:
a second gpu seems to be the only real working option.

I got mine to passthrough to a 2400g on my attempt at it, the then issue was code 43 because I used regular windows drivers instead of special ones you have to subscribe to or something

So idk about linux and VMs but straight windows works

1 Like

I posted the thread