Closed kevinkle closed 5 years ago
It's looking like it takes about all 62GB of ram + 118GB of swap to load 110 genomes
find . -maxdepth 1 -type f |head -1000|xargs cp -t "$destdir"
rule 'offset' found max_num_nodes to be 2344649
wontfix
. will likely switch to pytorch geomtric's implementation of diffpool instead of using diffpool directly
This seems to be max parallelism for 62.4G ram