Re: Dynamic symbol table
Lubos Lunak <[email protected]>
| Newsgroups | gmane.comp.kde.devel.optimize |
|---|---|
| Message-ID | <[email protected]> |
On Friday 16 of July 2004 17:57, Oswald Buddenhagen wrote: > On Fri, Jul 16, 2004 at 03:45:30PM +0200, Lubos Lunak wrote: > > And I did a test as well and it took somewhat shorter than demand > > paging the libs. Makes sense given that seeking is much slower than > > the actual reading. > > please do another test: instead of calling madvise, immediately read one > byte from each mapped page. i.e., don't let the kernel load the pages if > it thinks disks are idle, but enforce synchronous loading. i know for > sure that forced read-ahead improves memcpy behaviour more than cache > prefetch instructions. i wouldn't be surprised if a similar effect could > be observed with disk io - after all, with your current implementation > the program is running concurrently to madvise's effect and thus causes > trashing (theoretically). I think it'd be better if you tried that yourself, because I have no idea what you're talking about. How is synchronous page-by-page loading supposed to be faster than asynchronous madvise() ? -- Lubos Lunak KDE developer --------------------------------------------------------------------- SuSE CR, s.r.o. e-mail: [email protected] , [email protected] Drahobejlova 27 tel: +420 2 9654 2373 190 00 Praha 9 fax: +420 2 9654 2374 Czech Republic http://www.suse.cz/