Date: Sun, 26 Sep 2004 05:47:46 -0700 From: Ryan Freeman <ryan@slipgate.org> To: "Daniel O'Connor" <doconnor@gsoft.com.au> Cc: FreeBSD Mailing List <freebsd-current@freebsd.org> Subject: Re: Sound performance problems in FreeBSD Message-ID: <20040926124746.GA14197@slipgate.org> In-Reply-To: <200409262212.26882.doconnor@gsoft.com.au> References: <20040926025317.GA5812@slipgate.org> <200409261342.34874.doconnor@gsoft.com.au> <20040926114623.GA7990@slipgate.org> <200409262212.26882.doconnor@gsoft.com.au>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Sep 26, 2004 at 10:12:18PM +0930, Daniel O'Connor wrote: > On Sun, 26 Sep 2004 21:16, Ryan Freeman wrote: > > > I suspect that sysctl involves doing lots of locking and will do weird > > > things.. > > > Does it happen if you run a CPU using userland proc inseadt? > > > > The problem is, it isn't just that sysctl command that does it. I find that > > even running gkrellm makes the ticking worse. if i close gkrellm, it > > I'd say gkrellm probably pokes at various sysctl's to gather stats so it may > be a related problem. > > > reduces the ticks/sound stretches. that sysctl was just an easy way to make > > the problem _really_ stand out it seems. iirc when i was using 5.2.1 i > > found that just disabling the proc chart in gkrellm seemed to help a lot. > > haven't bothered to try it now. > > Does a purely CPU bound process cause the problem? > ie try something like.. > dd if=/dev/zero | md5 > well i gave that command a shot (i closed gkrellm first to reduce any clicking that may happen from it running) then started an mp3 with mpg123 alongside running that command. didn't skip a beat, so i guess its all related to sysctl... - ryan
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040926124746.GA14197>