GCN_Hybrid
2021-07-01 04:38:16 5 举报simple gcn flowchart
作者其他创作
大纲/内容
MPI_Scattervscatter Node data to other processes
yes
rank == 0
call \"create_graph\"
MPI_Init_thread(MPI_THREAD_FUNNELED)
start
end
compute local sum in each process
no
read Node data from files
call \"first_layer_transform\"
compute accuracy and print on the screen
receive Model data from rank 0
call \"second_layer_aggregate\"
broadcast Model data to other process
MPI_Allgathervsychronize Node data
create Nodes using for loop
read and load Model from file
call \"first_layer_aggregate\"
call \"second_layer_transform\"
收藏
0 条评论
下一页
