[Bioclusters] Re: BLAST job on SGE

Peiran Song peirans at cs.uoregon.edu
Fri Dec 3 15:18:04 EST 2004


Thanks Juan! I will try the configuration. Hope that could improve
the batch BLAST performance. When I tested a batch with 46 sequences, I
tried split the target database only, split query sequences only, and
split both, but no matter how, the cluster execution always takes longer
than the sequential execution. :( When split each sequence as one job, I
noticed that two jobs on one node executed for almost 13 mintues, while if
send them directly and almost simultaneously to one node, the execution
only takes 22 seconds...

Peiran




More information about the Bioclusters mailing list