Date: Mon, 14 Nov 2005 11:30:00 +0900 From: Pyun YongHyeon <pyunyh@gmail.com> To: "Matthew D. Fuller" <fullermd@over-yonder.net> Cc: Peter Jeremy <PeterJeremy@optushome.com.au>, freebsd-hackers@freebsd.org Subject: Re: poor fdc(4) performance Message-ID: <20051114023000.GB800@rndsoft.co.kr> In-Reply-To: <20051112161422.GB1485@over-yonder.net> References: <20051112071815.GD18405@rndsoft.co.kr> <20051112073711.GE39882@cirb503493.alcatel.com.au> <20051112074418.GE18405@rndsoft.co.kr> <20051112161422.GB1485@over-yonder.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Nov 12, 2005 at 10:14:22AM -0600, Matthew D. Fuller wrote: > On Sat, Nov 12, 2005 at 04:44:18PM +0900 I heard the voice of > Pyun YongHyeon, and lo! it spake thus: > > > > Yes, it could be. But I think the machine is fast enough to read > > sequential blocks. > > Try running it without SMP. There may be enough happening in the MP > locking bit that you end up falling behind. I remember noticing crap > fdc performance (without larger block sizes) on my dual PPro a little > while back. > Hmm, I think this is not the case as interrupt handler on fdc was registered with FAST handler. In addition, the handler just invokes wakeup(9), without any lock operations, to wake up kernel floppy worker thread. Note, I see the same issue on UP sparc64 too. -- Regards, Pyun YongHyeon
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20051114023000.GB800>