this might just be "luck" too. Linux pays no attention to how a program fits into memory and then into cache. I've seen significant performance changes in a program just based on it being run at different times, and getting a different set of real memory pages to run in.. sometimes this set of pages maps into the L2 cache (direct mapped) in a favorable way, the next time it maps into L2 in a poor way. Robert Hyatt Computer and Information Sciences [EMAIL PROTECTED] University of Alabama at Birmingham (205) 934-2213 115A Campbell Hall, UAB Station (205) 934-5473 FAX Birmingham, AL 35294-1170 On Sun, 22 Nov 1998, Emil Briggs wrote: > > > >I am intrigued too. There are vm enhancements in .36 so some paging patterns > >have no doubt improved but I assure you it doesnt sneak new improved FPU > >hardware into the pc > > > > I think the paging stuff could have a big effect on some numeric codes. > I've been working on optimizing code for the Origin2000 -- it has > hardware performance counters and you can see things like the amount of time > spent in TLB misses. You can run jobs and use different page sizes > for data, text and stack and look at the results and there is usually > an optimum set of values (I've seen 20% variations in performance on > some stuff that was memory intensive). > > > Regards > Emil >
