Parallel Jobs on Biowulf

make_swarmfile.R

make_swarmfile <- function(id_vector, scriptfile = "run.R", options = "--vanilla", outfile = "Rjobs"){
  
  runlist <- paste0("Rscript ", options, " ", scriptfile, " ", id_vector, " > run-", id_vector, ".Rout")
  cat(runlist, file = outfile, sep = "\n")
  
}

make_swarmfile(1:1000)

Rjobs

Rscript --vanilla run.R 1 > run-1.Rout
Rscript --vanilla run.R 2 > run-2.Rout
Rscript --vanilla run.R 3 > run-3.Rout
Rscript --vanilla run.R 4 > run-4.Rout
Rscript --vanilla run.R 5 > run-5.Rout
Rscript --vanilla run.R 6 > run-6.Rout
Rscript --vanilla run.R 7 > run-7.Rout
Rscript --vanilla run.R 8 > run-8.Rout
Rscript --vanilla run.R 9 > run-9.Rout
Rscript --vanilla run.R 10 > run-10.Rout

run.R

args <- commandArgs()
mean <- as.numeric(args[length(args)])
sample = rnorm(4444, mean)
saveRDS(sample, file = paste0("sample_", mean, ".RDS"))

submit

swarm -f Rjobs --module R

I ran into problems on a simulation and perused the user guide (http://biowulf.nih.gov/user_guide.html), and found:

**Note1: swarm will correctly allocate the correct number of processes to 2-, 4-, 16- or 24-core nodes.**

**Note2: if the programs you are running via swarm either (1) require more than 1 GB memory, or (2) are multi-threaded, you must specify additional information to swarm.**

See the swarm documentation (http://biowulf.nih.gov/apps/swarm.html)  for more details.

and found this.

-g #
--gb-per-process    gigabytes per process. By default swarm assumes each process will require 1 gb of        memory. Some applications require more than 1 gb, and so setting -g will restrict both the nodes allocated to the swarm and the number of processes per node to accomodate the memory requirement.

So I changed

swarm --autobundle -f Rjobs --module R 

to

swarm -g 4 --autobundle -f Rjobs --module R 

And it ...... WORKED!!!!!!!!!!!!

And when that didn't work, I

swarm -g 24 --autobundle -f Rjobs --module R 

and it worked. I didn't think -g 4 could go bigger than 4.