Create documentation on estimate hardware requirement for training GNN models on large graphs.

[jimjlu99]

📚 Documentation

[frozenbugs]

Da already have a workable script, assign to Da to post details to the issue.

[zheng-da]

https://gist.github.com/zheng-da/ca33af9048d09553f668f95bfd13cb0e

the code shows how to calculate the memory requirement for GNN training in DistDGL.

[TristonC]

Should this script be added into the DGL repo? @zheng-da

[jermainewang]

@TristonC I think it is too early at the moment. We are going to have a thorough review of the design of DistDGL soon due to the recent requests of reducing memory/startup overhead and supporting link prediction w/o information leakage. After that, we can have a more accurate estimation.

I will close this item. Feel free to reopen if you have further questions.