Hi Brandon > After a little more fiddling, it looks like the formula is > calculated at instance startup as: > > db_file_multiblock_read_count = db_cache_size/processes/9600, > with a max of .28% of the db_cache_size > > I'm not sure where that constant 9600 comes from, but I'd guess > it's based on db_block_size (8192 in my case) plus some other overhead. I don't know if it's helpful... Anyway, some years ago I investigated this issue. My conclusion, which was also published in my book (page 178), was the following: dfmbrc = min(1048576/db_block_size, db_cache_size/sessions/db_block_size) BTW, the dependency between db_cache_size and dfmbrc is, IMO, really bad... As a result, I never use this "auto-tune" feature. For this reason, back in 2005 I opened an SR about it. After some time the related bug was recognized to be a code bug. Few month ago (after almost 4 years of inactivity!) they changed the "status" to "FixBy->12". So, it might be that, at last, they'll solve the problem in the next major release... HTH Chris Antognini Troubleshooting Oracle Performance, Apress 2008 http://top.antognini.ch -- //www.freelists.org/webpage/oracle-l