search

BrendelGroup / xGDBvm

Code for the xGDBvm iPlant Atmosphere enabled genome annotation platform
http://brendelgroup.github.io/xGDBvm/
GNU General Public License v3.0
0 stars 3 forks source link

use of local CPU resources #10

Open vpbrendel opened 8 years ago

vpbrendel commented 8 years ago

The current setup does not use the full power of the local (VM) resources. In particular, GeneSeqer is always run in non-parallel mode, even on a VM with multiple processors.

Performance gains should be substantial by allowing the user an additional parameter when setting the Transcript Spliced Alignment options: "Number of processors". If set to a number >1 [default can be 1], then the script splitMakearryGSQ.pl (could be renamed in the process!) could be asked to invoke GeneSeqerMPI instead of GeneSeqer.

Suggestion: assign parameter passing to Jon, GeneSeqerMPI implementation to Volker, independent testing to Daniel

Priority: not urgent, but attractive

standage commented 8 years ago

Some sanity checking would be in order of course. If I were to test this feature, I would see how xGDBvm GUI handles the following as input.

It should be possible to programmatically probe the number of CPUs on the VM (/etc/procinfo?) and to make sure the number of requested processors does not exceed the number of available CPUs.

vpbrendel commented 8 years ago

Simple check in the script, using the following variable setting (echo included for your testing on your VM):

np=cat /proc/cpuinfo | grep processor | wc -l; echo $np