| |||||||||||||
Protected RAMFSProtected RAMFSPosted Jun 25, 2009 15:01 UTC (Thu) by nix (subscriber, #2304)Parent article: Protected RAMFS
With all those pagetable changes and TLB flushes I'm surprised that performance is as high as 13Mb/s. 26 million pagetable changes per second is a hell of a lot more than I thought we could manage even on a fast system.
(Log in to post comments)
Protected RAMFS Posted Jun 25, 2009 15:09 UTC (Thu) by jake (editor, #205) [Link]
> 26 million pagetable changes per second
I don't think it does 2 page table changes per *byte*, presumably per page, which is still rather a lot.
jake
Protected RAMFS Posted Jun 25, 2009 19:05 UTC (Thu) by nix (subscriber, #2304) [Link]
Ah, right: I was stuck behind a stupid corporate firewall at the time so
couldn't check the code. I was assuming a pagetable change on every write(), but I guess they could amortize it to some extent.
|
|||||||||||||