MAKER parallelizes at multiple levels. For the ab initio predictors, it will run multiple contigs simultaneously (so each one will get their own ab initio predictor running). For large contigs it will further divide it into 10Mb chunks, and each will run simultaneously.
On Sep 26, 2017, at 1:23 AM, Xabier Vázquez-Campos <[hidden email]> wrote:
We finally got Maker working with MPI (mpich, openmpi was a dead end...) and I have a question about how Maker distributes the computation load.
I know, correct me if I'm wrong, that with MPI, Maker runs blast in parallel (1 instance per thread) for protein2genome and est2genome. This indeed improves enormously the speed for the initial run.
But, does it take advance of this at the time of running the gene predictors? I think there is no benefit on multiple cpus in non-MPI mode but I have no idea in MPI.
Thank you in advance,
Xabier Vázquez-Campos, PhD Research Associate NSW Systems Biology Initiative School of Biotechnology and Biomolecular Sciences
The University of New South Wales Sydney NSW 2052 AUSTRALIA