From owner-freebsd-alpha Thu Feb 13 4:13: 2 2003 Delivered-To: freebsd-alpha@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 6479737B401 for ; Thu, 13 Feb 2003 04:12:59 -0800 (PST) Received: from smtp.web.de (smtp02.web.de [217.72.192.151]) by mx1.FreeBSD.org (Postfix) with ESMTP id A117F43F85 for ; Thu, 13 Feb 2003 04:12:57 -0800 (PST) (envelope-from Jan.Lentfer@web.de) Received: from [213.157.1.107] (helo=floundjan.homeip.net) by smtp.web.de with esmtp (WEB.DE(Exim) 4.95 #31) id 18jIEM-0004NE-00; Thu, 13 Feb 2003 13:12:42 +0100 Received: from localhost (localhost.lan [127.0.0.1]) by floundjan.homeip.net (Postfix on FreeBSD 4.5) with ESMTP id 4CEC11C3; Thu, 13 Feb 2003 13:12:41 +0100 (CET) Received: from localhost.localdomain (neslonek.bio.tu-darmstadt.de [130.83.207.147]) by floundjan.homeip.net (Postfix on FreeBSD 4.5) with ESMTP id C509BFF; Thu, 13 Feb 2003 13:12:35 +0100 (CET) Subject: Re: Why is XFree86 on FreeBSD/Alpha much slower? From: Jan Lentfer To: Marco Beishuizen Cc: Andrew Gallatin , FreeBSD alpha mailing list In-Reply-To: References: <15945.23721.759825.313603@grasshopper.cs.duke.edu> <15945.24538.794799.926276@grasshopper.cs.duke.edu> <15945.25465.768071.583030@grasshopper.cs.duke.edu> <15945.26349.357142.151047@grasshopper.cs.duke.edu> <15945.27540.655333.324376@grasshopper.cs.duke.edu> Content-Type: text/plain Content-Transfer-Encoding: 7bit X-Mailer: Ximian Evolution 1.0.5 Date: 13 Feb 2003 13:14:15 +0100 Message-Id: <1045138456.618.5.camel@neslonek> Mime-Version: 1.0 X-Virus-Scanned: by AMaViS perl-10 Sender: owner-freebsd-alpha@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org Am Die, 2003-02-11 um 23.38 schrieb Marco Beishuizen: > On Tue, 11 Feb 2003, the wise Andrew Gallatin spoke, and said: > > > > > Search for "set pci_device_override" at > > > > http://www.freebsd.org/releases/5.0R/hardware-alpha.html > > > > The bug in question occurrs when DMAing across a page boundary. > > In my experience, this bug causes the machine to lock solid, not to > > exhibit corruption. > > > > Like Fred said, its a graphics controller not a scsi adapter. Give it > > a shot. > > Well, the Matrox is in a 64 bit slot now, and I have to say it's a big > improvement! I did a 'set pci_device_override -1' in the console, and > graphics performance in acrobat and gimp is a lot faster now. > > Thank you all for the tips. Hi all, I tried to gather some numbers so one can compare the graphic performance of 32bit vs. 64bit slot. I did "x11perf -all" on a PWS 500a, Debian Linux 2.4.20, Xfree 4.2.1, ATI Radeon with DRI. Here are the results from x11perfcomp: lentferj@neslonek:~$ x11perfcomp -r ati32bit ati64bit 1: ati32bit 2: ati64bit 1 2 Operation -------- ----------------- ----------------- 2140000.0 4880000.0 ( 2.28) Dot 1170000.0 1190000.0 ( 1.02) 1x1 rectangle 1170000.0 1180000.0 ( 1.01) 10x10 rectangle 32700.0 32700.0 ( 1.00) 100x100 rectangle 1520.0 1530.0 ( 1.01) 500x500 rectangle 691000.0 715000.0 ( 1.03) 1x1 stippled rectangle (8x8 stipple) 633000.0 632000.0 ( 1.00) 10x10 stippled rectangle (8x8 stipple) 10500.0 10900.0 ( 1.04) 100x100 stippled rectangle (8x8 stipple) 350.0 352.0 ( 1.01) 500x500 stippled rectangle (8x8 stipple) 683000.0 714000.0 ( 1.05) 1x1 opaque stippled rectangle (8x8 stipple) 673000.0 690000.0 ( 1.03) 10x10 opaque stippled rectangle (8x8 stipple) 18400.0 18000.0 ( 0.98) 100x100 opaque stippled rectangle (8x8 stipple) 839.0 749.0 ( 0.89) 500x500 opaque stippled rectangle (8x8 stipple) 678000.0 696000.0 ( 1.03) 1x1 tiled rectangle (4x4 tile) 667000.0 685000.0 ( 1.03) 10x10 tiled rectangle (4x4 tile) 18400.0 18100.0 ( 0.98) 100x100 tiled rectangle (4x4 tile) 748.0 749.0 ( 1.00) 500x500 tiled rectangle (4x4 tile) 709000.0 720000.0 ( 1.02) 1x1 stippled rectangle (17x15 stipple) 664000.0 674000.0 ( 1.02) 10x10 stippled rectangle (17x15 stipple) 11000.0 11000.0 ( 1.00) 100x100 stippled rectangle (17x15 stipple) 363.0 363.0 ( 1.00) 500x500 stippled rectangle (17x15 stipple) 709000.0 713000.0 ( 1.01) 1x1 opaque stippled rectangle (17x15 stipple) 679000.0 700000.0 ( 1.03) 10x10 opaque stippled rectangle (17x15 stipple) 18400.0 18500.0 ( 1.01) 100x100 opaque stippled rectangle (17x15 stipple) 746.0 785.0 ( 1.05) 500x500 opaque stippled rectangle (17x15 stipple) 708000.0 724000.0 ( 1.02) 1x1 tiled rectangle (17x15 tile) 679000.0 698000.0 ( 1.03) 10x10 tiled rectangle (17x15 tile) 18300.0 18500.0 ( 1.01) 100x100 tiled rectangle (17x15 tile) 731.0 785.0 ( 1.07) 500x500 tiled rectangle (17x15 tile) 737000.0 752000.0 ( 1.02) 1x1 stippled rectangle (161x145 stipple) 692000.0 692000.0 ( 1.00) 10x10 stippled rectangle (161x145 stipple) 12800.0 12500.0 ( 0.98) 100x100 stippled rectangle (161x145 stipple) 400.0 407.0 ( 1.02) 500x500 stippled rectangle (161x145 stipple) 740000.0 755000.0 ( 1.02) 1x1 opaque stippled rectangle (161x145 stipple) 721000.0 731000.0 ( 1.01) 10x10 opaque stippled rectangle (161x145 stipple) 18900.0 19300.0 ( 1.02) 100x100 opaque stippled rectangle (161x145 stipple) 805.0 847.0 ( 1.05) 500x500 opaque stippled rectangle (161x145 stipple) 740000.0 754000.0 ( 1.02) 1x1 tiled rectangle (161x145 tile) 720000.0 737000.0 ( 1.02) 10x10 tiled rectangle (161x145 tile) 18900.0 18300.0 ( 0.97) 100x100 tiled rectangle (161x145 tile) 805.0 788.0 ( 0.98) 500x500 tiled rectangle (161x145 tile) 739000.0 755000.0 ( 1.02) 1x1 tiled rectangle (216x208 tile) 725000.0 742000.0 ( 1.02) 10x10 tiled rectangle (216x208 tile) 18300.0 17600.0 ( 0.96) 100x100 tiled rectangle (216x208 tile) 808.0 782.0 ( 0.97) 500x500 tiled rectangle (216x208 tile) 1080000.0 1160000.0 ( 1.07) 1-pixel line segment 1240000.0 1280000.0 ( 1.03) 10-pixel line segment 427000.0 428000.0 ( 1.00) 100-pixel line segment 86300.0 86500.0 ( 1.00) 500-pixel line segment 424000.0 439000.0 ( 1.04) 100-pixel line segment (1 kid) 328000.0 352000.0 ( 1.07) 100-pixel line segment (2 kids) I put the graphic card in the lower one of the two 64bit slot. Would it make any difference it I put in the upper of the two? Cheers, Jan To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-alpha" in the body of the message