search

glennhickey / progressiveCactus

Distribution package for the Prgressive Cactus multiple genome aligner. Dependencies are linked as submodules
Other
79 stars 26 forks source link

HowTo evaluate progressiveCactus' use of available computational resources? #89

Closed malcook closed 6 years ago

malcook commented 6 years ago

When running progressive cactus on a large memory (.5 T) 40-core running CentOS7 I witness long stretches of almost 0% CPU utilization. Occasionally I see a brief CPU spike in one of: cactus_bar cPecanRealign ktserver. The cactus.log has quite a bit of timing stats, etc, but I really don't know how to interpret it. One thing that seems ominous is the presence of multiple "Blocking on ktserver" message in the log. How should I interpret the CPU utilization and the log entries in light of PC's architecture, and possibly discover opportunities for better resource allocation? FWIW: I also have a Open Grid Engine cluster available to me with about 50 16GB 8-core nodes.

malcook commented 6 years ago

Closing this myself after closer reading of Computation Time & Memory Usage which provides good detail. I am still somewhat befuddled by long stretches of 0% CPU... I don't think I am I/O bound....