vGPU Support

asked 2018-03-15 05:11:04 -0500

Dear Openstack members,

I am working on a feature that needs to implement vGPU in Openstack platform. We use Redhat OSP10 (based on Newton). After going through numerous pages and blogs, i am able to understand that i can use a whole GPU using pcipassthrough functionality available in Openstack nova. I have not tested or verified it yet. But i am assuming it will work. However, we are planning to make use of vGPU feature offerred by nVidia GPU's and want to provision openstack instances with a specific profile (1GB or 2GB or 4GB or 8GB or 16GB). Any ideas, as to which Openstack release support vGPU? I understand that vGPU functionality has to be enabled in KVM and libvirt libraries first so that it is available in openstack nova. I have been reading this following thread ( ( and it seems some work is already done for libvirt and KVM's.

Many thanks for your time and suggestions.

warm regards Nebu

edit retag flag offensive close merge delete