Betty: Enabling Large Scale GNN Training with Batch Level Graph Partitioning

This artifact includes the source codes and expected experimental data for replicating the evaluations in this paper.
We implement figure 2 to denote the OOM situation of current advanced GNN training, and applied figure 10 to illustrate Betty break the memory wall. We implement memory consumption estimation during the workflow of Betty, shown in figure 5. We use figure 12 to denote the tendency of peak memory consumption and training time per epoch as the number of micro batches increases. And the model convergence is not impacted by Betty and micro-batch training can be proved by the figure 13.
The framework of Betty is developed upon DGL(pytorch backend). The requirements: pytorch >= 1.7, DGL >= 0.7. The other software dependency include sortedcontainers, pyvis, pynvml, tqdm, pymetis, seaborn.
Our experiments result denoted in paper were collected from the machine with a RTX6000 GPU(24 GB memory) and Intel(R) Xeon(R) Gold 6126 CPU @ 2.60GHz. You can use a different configuration with at least one GPU.