Which codes can repoduce the Figure4 Runtime overhead in paper?

[LukeLIN-web]

I wonder how we can evaluate Figure 4, Table4 and 5 in the paper: Runtime overhead in relation to the inter-/intra-
connectivity ratio of mini-batches, both for serial and concurrent history access patterns.
I am sorry that I cannot find it in the repo.

[rusty1s]

You are right that this part is not included in this repo, I am sorry.

You should be able to run "serial" mode by setting _async=False in the ScalableGNN base class.