Download NC_006511 sequence from remote DB (~10 Mb).
Make >30 copies of the file with different names.
Create a scheme to perform the HMM search. Set all the copies as an input file list. Run the search.
In my case it consumed > 500 mb when it was passing the 20th file. It seems to be a memory leak or bad memory management in case of multiple files. It must allocate memory for one file to performe the search not for a banch o them.