Bsoft was developed by groups at the University of Basel, NIAMS, and Caltech.
The Bsoft programs have standard Unix command-line syntax, and so are ideally suited to running as a batch job or as a swarm of batch jobs.
Sample Bsoft batch script:
#!/bin/bash module load Bsoft cd /data/user/myjob bimg -verbose 7 -truncate 0,100 input.img output.img
This job would be submitted with
This job would be submitted to a single core (2 hyperthreaded CPUs) and 4 GB of memory. If more than the default 4 GB of memory is required, you would submit with:
sbatch --mem=#g jobscriptwhere '#' is the number of GigaBytes of memory required.
Biowulf users will typically want to process large numbers of image files via Bsoft. This is most easily done using the swarm utility.
Now set up a swarm command file along the following lines:
# this file is called swarmfile bimg -verbose 7 -truncate 0,100 input1.img output1.img bimg -verbose 7 -truncate 0,100 input2.img output2.img bimg -verbose 7 -truncate 0,100 input3.img output3.img bimg -verbose 7 -truncate 0,100 input4.img output4.img bimg -verbose 7 -truncate 0,100 input5.img output5.img [...]
Submit this with
biowulf% swarm -f swarmfile --module Bsoft/2.0.0
If each Bsoft process in the swarm command file requires more than 4 GB of memory, use
biowulf% swarm -g # -f swarmfile --module Bsoft/1.9.0
To run Bsoft interactively on Biowulf, allocate an interactive session and run the process there.
biowulf% sinteractive cn0203% module load bsoft cn0203% ...bsoft commands ... cn0203% exit