Hongzhi (Steve), Chen

functional, subgraph are done. convert is pending.

As discussed with @peizhou001 , it is done.

https://github.com/dmlc/dgl/issues/4860

@peizhou001 to clarify the problem, many graph, or single graph. Let's decide the plan later.

Sorry for the late response. @nv-dlasalle can you help us understand your use case here? We are redesigning the Dataloader, we may simplifier the DGLBlocker. If this happens, the problem...

As graphbolt has released in 2024, can you take a look at https://docs.dgl.ai/stochastic_training/link_prediction.html to see whether it solves your issue. Feel free to reopen if the problem still persist.

The code is here, can you create a PR? docs/source/guide/minibatch-gpu-sampling.rst

Let's leave this as draft, and merge it once the end to end code is ready to clean up.

> > > > How much slower is this new implementation compared to the old one? > > > > > > > > > @mfbalin The time complexity should...

The code there is just for demonstration, which is not expected to be runnable seamlessly, can you try https://github.com/dmlc/dgl/tree/master/examples/distributed/graphsage.