Date: Fri, 26 Oct 2001 20:21:29 +0200 From: Poul-Henning Kamp <phk@critter.freebsd.dk> To: Alfred Perlstein <bright@mu.org> Cc: Julian Elischer <julian@elischer.org>, Terry Lambert <tlambert2@mindspring.com>, Peter Wemm <peter@wemm.org>, arch@FreeBSD.ORG Subject: Re: 64 bit times revisited.. Message-ID: <5201.1004120489@critter.freebsd.dk> In-Reply-To: Your message of "Fri, 26 Oct 2001 12:35:26 CDT." <20011026123526.F15052@elvis.mu.org>
next in thread | previous in thread | raw e-mail | index | archive | help
In message <20011026123526.F15052@elvis.mu.org>, Alfred Perlstein writes: >> Because applications in core write files on disk ? :-) > >Excuse me for being daft, but isn't some of the resolution currently >in the on-disk portion more than the time it takes to write and >re-read the data from most media? Are you concerned with faster >media? Perhaps MFS? Not all filesystems are backed by disk, some (like DEVFS) have no physical backing at all. Second, with softupdates, even physically backed disks are not constrained to a physical movement for each transaction. Third, I want people to consider for a moment what the world looked like 10 years ago. Now who here dare predict if we will need or not need nanosecond (or better) resolution for timestamps 10 years down the road ? After all, PC cpu clock frequencies have only increased by a factor 42[*] or so in the last 10 years... Ten years down the road, people might curse us for wasting two bits which could have split the nanoseconds to solve a problem which is by then still 27 years down the road... So what I'm advocating is to discuss this with Kirk for his FFS2 layout, and realize that very few of the media which holds todays filesystems will still be rotating 10 years from now. I do, by the way, find it much more interesting to discuss if future inodes should contain counters counting number of R/O opens, number of R/W opens, number of sectors written etc etc. Such statistics could be used to optimize layout based on actual statistics, rather than statistical guesswork as we do now... Poul-Henning [*] 1400MHz / 33 MHz = 42.4242424242, there is probably no cosmic significance in this numer. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 phk@FreeBSD.ORG | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-arch" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5201.1004120489>