It would be nice if there was a separate server executable you could launch on something like RunPod with a H100 GPU. Then you could run the interface locally as always but send any generation jobs to the server rather than running them locally on much slower hardware.
To do this on a Windows PC, create a Virtual Machine, load Windows and TVAI and select the link for the provider’s online GPU as the Virtual Machine’s GPU. That will create a separate instance of TVAI in its own environment that uses the cloud GPU as if it’s local HW.
1 Like
What do you mean by this? I am not aware of any software that can do that.
When you create a VM in Windows, you configure it just like you would a physical one, selecting virtual GPUs, drives, network adapters, etc. In most cases, you just select the physical ones in your own computer, but you can also specify cloud-based substitutes.
Windows VM:
Cloud GPUs:
Maybe start on BOINC for getting done jobs via cloud computing?
I think most modern cloud computing schemes require a provider that is actually running the app on its hardware. The folks at Topaz did mention the possibility of some kind of farm scheme, but IMO the prospects of a commercial SW product using a public access distributed computing system seems doubtful.
