gistfile1.txt

For that step (match), the software first reads into memory a binary version of the reference
genome:

/stornext/snfs4/next-gen/solid/bf.references/h/hsap.36.1.hg18/hsap_36.1_hg18.fa.nt.brg

Then it splits the input data (reads from stornext) into 8 tmp files (/space1/tmp).

Then per each of the indexes (13G files located in
/stornext/snfs4/next-gen/solid/bf.references/h/hsap.36.1.hg18/hsap_36.1_hg18.fa.cs.*.bif) 
loads one a time and spawns 8 threads each processing the data from the tmp files (8 files).
The output is also in space1/tmp.

When the software has processes all the 10 indexes. It reads the intermediate results 
files (space1/tmp) and dumps the final results in stornext.

Here you have a cpu profiling of that step with different versions of the software:

http://is04607.com/bfast/bf.top/

Hope that helps,
-drd