multicore performance issue

[ramcn]

Hello,

I am running graphmap align on a multicore xeon server on an nanopore dataset single flow cell rel3-nanopore-wgs-152889212-FAB45271.fastq.gz. The server has 2 sockets, with 14 cores each and hyper threading enabled. So it can run upto 56threads in parallel.

Despite using "-t 28" the performance of graphmap is very slow. It takes nearly 10hrs to complete where as other tool like ngmlr can complete in 30mins. When I monitor the system load during graphmap alignment, the CPU utilization is very low and only 2 of the 56 cores show 100% utilization.

Is this a known issue with graphmap. I can provide more performance insights if required.

Thank you, Ram