Beginner question - how do apps make use of a GPU?

Apps and Virtualization
, ,
1

Hello all

I’m using TrueNAS Scale Dragonfish-24.04.2 running on an HP microserver (gen8).

I have recently added an MSI GeForce GT 1030, for use with the following TrueNAS apps:

Frigate
MotionEye
Jellyfin

for help with decoding and transcoding.

  1. If, in each of the apps I specify support for one Nvidia GPU, does the first app to be run grab the GPU resource, with the remaining two being excluded?

  2. How would the single GPU resource get shared between the three when I run more than one app?

Thanks in advance for any simple guidance you can point me towards.

EB

1 Like
2

The GPU gets “sliced” into parts. The maximum number of slices for a nvidia gpu is 5. When you assign a gpu to an app it grabs one of those slices and resources are shared between all slices.

3

Perfect - yes, the slicing sharing multiplexing approach; that’s easy for me to understand. Thanks Lars.