What are the best workloads for TFGrid using GPU

We believe it would be awesome to support GPU’ on TFgrid, a lot of them will become available:


What would the best applications be for GPU?


I think one application for GPU would be to host a server that is almost like a VPS, where people can use it to stream games that their computer can not run at home. If anyone knows how to set this up, I would like to be able to do this on my node.

1 Like

tensorflow machine learning… research institutions and universities can now have the benefit of local capacity (available for research etc) and overflow capacity (when needed) in the TF Grid with minimum admin overhead and focus on the end goal, not the road towards getting there…


GPU application on TF grid would be good for end-to-end data science workflows using RAPIDS™ open source software libraries using local capacity close to where its needed.


It would be nice if workloads of folding@home could be deployed on the grid as these support GPU.

Folding @ Home workloads are executed free of charge, but it would be nice if you could donate part of your node capacity to Folding@home tasks or some other clever thing that works.

The science community using Folding@home is also a very nice community to make familiar with Threefold.


https://foldingathome.org/, nice indeed, we would have to use our DAO to make sure it can run for free on machines who have the capacity if farmers agree.


I’ve found thetatoken.org to be one of the best decentralized applications for cloud GPU utilization:
Streaming, video on demand, GPU-rendering jobs, machine learning, etc.

They are partnered with foldingathome as well.
Possible to run Theta-nodes and/or workloads on the TF-grid.