1
1
I have several large analysis jobs in parallel that run over the same couple hundred gigabytes of data that are i/o bound. When these run the Superfetch cache quickly consumes all available free memory, however I suspect the jobs are not synchronous enough in terms of data locality and the cache is just thrashing.
Is there a way to get any deeper information about exactly what the hit rates for the Superfetch cache are?
Chuu
Posted 2012-08-31T19:22:32.433
Reputation: 549