Leyuan Wang
Leyuan Wang
@jowens I don't think this affects correctness. Else condition here is trying to get more related nodes pruned. If the loading `subgraphs[src] >= constrain[0]` happens before the writing `atomicAdd(subgraphs +...
To avoid the data race, we can use double buffering method. Say in odd iterations, we do loading from buffer1 and writing to buffer2, and in even iterations, we do...
> And specifically this is on the array `subgraphs`? Yeah, that's right.
@nvpohanh Just curious what kind of cooling method do you use for T4.