Date: Thu, 24 Apr 1997 13:18:33 -0700 From: David Greenman <dg@root.com> To: Poul-Henning Kamp <phk@dk.tfs.com> Cc: fs@freebsd.org Subject: Re: the namei cache... Message-ID: <199704242018.NAA10732@root.com> In-Reply-To: Your message of "Thu, 24 Apr 1997 21:40:49 %2B0200." <1303.861910849@critter>
next in thread | previous in thread | raw e-mail | index | archive | help
>In message <199704241934.MAA10488@root.com>, David Greenman writes: >>>about 1/4th of the additions. Getting reed of the div wouldn't also be bad >>>but might not be worth it. >> >> Not worth it? At greater than 50 cycles, the divide is more expensive >>than the entire add loop. > >But loosing the hash entirely is >even< faster, right ? :-) Maybe. Here's the problem with your approach as I see it: On wcarchive, we have about 44,000 files in the namei cache. The hash table size is approximately 44,000 entries large (rounded up or down to the nearest prime...I forget). This means that there is an average of about 1 file in each hash bucket...or very little to search through. Of course there are far fewer than 1 directory per file, so now if you change this so that all of the files for a directory are on one list, then you're going to have to do a lot more work to find it. I think the average directory length on wcarchive is about 50 files or so, so this would be about 50 times more entries in the list compared to the old hash table scheme. -DG David Greenman Core-team/Principal Architect, The FreeBSD Project
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199704242018.NAA10732>