DGL Support for CUDA 12.2 and PyTorch 2.1


I recently updated my server to CUDA 12.2 and PyTorch 2.1. However, I’ve encountered compatibility issues with DGL, as it seems not to support these versions. Could you please provide some information or guidance on the current status of DGL’s support for CUDA 12.X and PyTorch 2.X?

Is there a timeline for when DGL is expected to support these newer versions, or are there any workarounds or recommendations for users in a similar situation?

We’re planing to roll out new DGL release next month(2023.11) and torch 2.1(with cuda 12.2) will be supported.

As a workaround for now, you could try to build DGL from source(master or 1.1.x branch) with torch 2.1 on your own. see guide here.

1 Like

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.