It is at the ranking/sorting stage that should parallelise
[ INFO ][ 2019-03-20T16:36:05 ][ aggregate ]: Done aggregating. Ranking and standardizing aggregated edges
This could have been run on a version that was compiled without PSTL. Trying to reproduce this has been a bit tricky. I'll keep looking but so far it looks like we had a compilation issue, rather than a bug
Allocation is: NumNodes=1 NumCPUs=32 NumTasks=32 CPUs/Task=1 ReqB:S:C:T=0:0:: TRES=cpu=32,mem=128G,node=1
Code is: CPU=32 export OMP_NUM_THREADS=$CPU seidr aggregate -m irp -k -O $CPU $@
It is at the ranking/sorting stage that should parallelise [ INFO ][ 2019-03-20T16:36:05 ][ aggregate ]: Done aggregating. Ranking and standardizing aggregated edges
But it only uses 1 CPU 100%