Parallel CLI

If you are using an existing third party application to process your data, you might already have parallel computing available to you. You would have to specify it on the command-line. The most common forms of the ‘-p’, ‘-np’, or ‘–parallel’ options. These options are almost always followed by a number indicating a processor count:

sort --parallel 8 ...

Another example might be:

bowtie2 -p 4 ...

Some applications, however, allows parallel computing but does not require specification of processors on the command-line. They are controlled by an environment variable ‘OMP_NUM_THREADS’. These applications are utilizing the openMP library - and it is a very common paradigm to add parallel computing to applications:

export OMP_NUM_THREADS=8

Gromacs, a popular molecular dynamic program, is controlled this way.

Tip

Check the documenation of your application, specifically looking for sections outlining parallel computing. It’s important to be able to recognize what type of parallel computing your application offers. If your program uses OMP_NUM_THREADS, or uses -p, -np, etc... to specify processor number - then it almost certainly uses shared parallel computing. And therefore, it will not be adequate to alleviate memory usage limitations.