[ICLR 2022] "PipeGCN: Efficient Full-Graph Training of Graph Convolutional Networks with Pipelined Feature Communication" by Cheng Wan, Youjie Li, Cameron R. Wolfe, Anastasios Kyrillidis, Nam Sung Kim, Yingyan Lin
Hi,
I breifly read the code, and have no idea how the model get the buffered data. The implementation of GraphSAGELayer directly use the graph function provided by DGL. I guess there might be some cusomtization in DGL, but did not find it easily. May be there are some other aspects that I did not notice.