Does Gaudi support CUDA?

Training FAQ
1

CUDA is for NVIDIA GPU. The best way to use Gaudi is to use a framework that is built for Gaudi. Currently, TensorFlow and PyTorch are integrated with SynapseAI and optimized for Gaudi.

In cases where a user would need to create custom kernels, Habana provides an SDK for creating kernels. Users can start on our Developer Site for custom kernels to access the tools, documentation and videos.

Additionally, In TensorFlow, you can write custom Ops. For more information, visit Custom OP example on our GitHub page.

2

When following the “custom op” link, I get a “page not found”…

3

The custom op leads to examples of writing TPC code for TF. However we have dropped support for Tensorflow. You can still find the TF page on release 1.14:

You can find more TPC resources here:

https://docs.habana.ai/en/latest/TPC/index.html