Date: Wed, 2 Jan 2008 11:49:47 -0500 From: "Josh Carroll" <josh.carroll@gmail.com> To: "Anish Mistry" <mistry.7@osu.edu> Cc: Kip Macy <kip.macy@gmail.com>, Kris Kennaway <kris@freebsd.org>, freebsd-stable@freebsd.org, "\[LoN\]Kamikaze" <LoN_Kamikaze@gmx.de> Subject: Re: RELENG_7 jerky mouse and skipping sound (still a problem -BETA3) Message-ID: <8cb6106e0801020849n6f504a05wddd315c663932bc5@mail.gmail.com> In-Reply-To: <200801020136.54856.mistry.7@osu.edu> References: <200710171228.39123.mistry.7@osu.edu> <b1fa29170711191022k2e3ea2afocdcd32ca0279c41a@mail.gmail.com> <4777942D.9010209@FreeBSD.org> <200801020136.54856.mistry.7@osu.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
> > Can you confirm that this fix helped for you? i.e. do you still > > see the problem? > FreeBSD 7.0-RC1 #14: Sun Dec 30 21:50:59 EST 2007 > I'm still seeing this problem, but it isn't nearly as bad. I still > get some jerky mouse movement, but music doesn't skip now when I'm > compiling. I noticed severe sluggishness in X.org the other day when I logged into the console on my 7.0-RC1 box. With firefox open, the mouse was very unresponsive, and at one point I counted and it took ~10 seconds to register a mouse click in firefox. The mouse jerkiness I fixed by starting moused and using /dev/sysmouse instead of /dev/psm0. However, the sluggishness overall in X was still there, and firefox took a lot longer than it should to render pages. I then noticed in my xorg.conf, I had the Depth set to 16-bit (this is on the X.org nv driver). I switched to 24-bit and it is MUCH faster. I ran the wm_torture (http://www.rasterman.com/files/wm_torture-0.1.tar.gz) response test for both 16-bit and 24-bit, and here are the results: 16-bit: Test: map_response MIN: 0.042016s, MAX: 0.046576, AVG: 0.044307 24-bit: Test: map_response MIN: 0.002540s, MAX: 0.005877, AVG: 0.004306 That's over a 10x speed up. I believe those numbers are saying with 16-bit it took an average of 44ms to draw a window, while at 24-bit it took 4.3ms. So, for anyone using the nv (perhaps this applies to other cards/drivers?), check that you are not using 16-bit color depth, as it really seems to hinder performance. Note: this is on an 7.0-RC1 box (running ULE), so the ithread inversion has already been included in the kernel I'm running, so that may have helped as well. But I did not need to set FULL_PREEMPTION or HZ to achieve the good performance. Thanks, Josh
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?8cb6106e0801020849n6f504a05wddd315c663932bc5>