primeMemory seemingly not being called
I've run into a situation where a customer is running lastal5 on a cluster I operate against a lustre file system. The performance is abysmal largely because of the read pattern against the reference data set. I've figured out the user can just prepend the job run with a set of dd's that read the reference dataset into /dev/null, thus populating the node's page cache. Runs from then are a well within the expected bounds. I've noticed that src/fileMap.cc's primeMemory() does exactly this but does't seem to get called by the user's lastal5 run. Is this expected or a bug?