From owner-freebsd-x11@FreeBSD.ORG Wed Sep 12 08:56:54 2012 Return-Path: Delivered-To: freebsd-x11@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 4F5A61065674 for ; Wed, 12 Sep 2012 08:56:54 +0000 (UTC) (envelope-from l.pizzamiglio@bally-wulff.de) Received: from mail.bally-wulff.de (mail.bally-wulff.de [212.144.118.8]) by mx1.freebsd.org (Postfix) with ESMTP id E84268FC19 for ; Wed, 12 Sep 2012 08:56:53 +0000 (UTC) Received: from bwex.bally-wulff.de (bwex.bally-wulff.de [192.168.204.106]) by mail.bally-wulff.de (Postfix) with ESMTP id D6D664E076; Wed, 12 Sep 2012 10:49:50 +0200 (CEST) Received: from pizzamig.bally.de ([192.168.205.30]) by bwex.bally-wulff.de with Microsoft SMTPSVC(6.0.3790.4675); Wed, 12 Sep 2012 10:50:29 +0200 Message-ID: <50504CD5.1070202@bally-wulff.de> Date: Wed, 12 Sep 2012 10:50:29 +0200 From: Luca Pizzamiglio User-Agent: Mozilla/5.0 (X11; FreeBSD i386; rv:15.0) Gecko/20120911 Thunderbird/15.0.1 MIME-Version: 1.0 To: "Yuri K. Shatroff" References: <504F4EEC.6070300@yandex.ru> <504FA9DA.2080402@yandex.ru> In-Reply-To: <504FA9DA.2080402@yandex.ru> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 12 Sep 2012 08:50:29.0310 (UTC) FILETIME=[A91105E0:01CD90C3] Cc: freebsd-x11@freebsd.org Subject: Re: Ivy Bridge and Xorg 7.7 on i386 FreeBSD 9-1BETA: no hardware acceleration? X-BeenThere: freebsd-x11@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: X11 on FreeBSD -- maintaining and support List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 12 Sep 2012 08:56:54 -0000 On 09/11/12 23:15, Yuri K. Shatroff wrote: > On 11.09.2012 19:24, archibald wrote: >> On 2012-09-11 15:47, Yuri K. Shatroff wrote: >>> Greetings x11@, >>> >>> I've got the same problem using Ivy Bridge HD4000 (with notorious >>> optimus). The DE (KDE4) is severely lagging and experiencing seconds' >>> delays in reaction to mouse/kbd events. So using the desktop is >>> practically impossible. In `top`, the Xorg process consumes up to 100% >>> CPU. >>> When I switch to VESA driver, there is no DE performance problem >>> (rendering and reactions almost instant, CPU usage by Xorg about >>> 6-10%). >>> I have exactly the same software versions installed on another PC but >>> with nvidia hardware and x11/nvidia-driver, and everything works there >>> fine, too. So I assume the problem is somewhere around drivers (either >>> xf86-video-intel, or Mesa layer, or the i1915kms module). >>> >>> My system is: >>> FreeBSD 9.1-PRERELEASE (GENERIC) #0 r240286: Mon Sep 10 14:27:33 MSK >>> 2012 >>> the latest Xorg from the xorg-dev trunk (xorg-server 1.12.4, >>> xf86-video-intel-2.20.4) >>> WITH_KMS=true >>> WITH_NEW_XORG=true >> >> What is the output of: >> $ xdriinfo >> and >> $ glxinfo | grep OpenGL >> >> (glxinfo is found in graphics/mesa-demos, xdriinfo is in x11/xdriinfo) >> >> archibald > > $ xdriinfo > Screen 0: i965 > $ glxinfo | grep OpenGL > OpenGL vendor string: Tungsten Graphics, Inc > OpenGL renderer string: Mesa DRI Intel(R) Ivybridge Mobile > OpenGL version string: 2.1 Mesa 8.0.4 > OpenGL shading language version string: 1.30 > OpenGL extensions: > > I'd be glad if that helps, looks like nothing special to me though > Hi Yuri, I tested KMS and NEW_XORG with BSD9 on SandyBridge and IvyBridge. I've spoken with @x11 guys on IRC and it seems that the problem is: a bit/knob is not recognized/enabled/activated. The problem could be everywhere: kernel driver, intel DDX driver, libdrm, mesa (libdri-libGL-ecc), ecc. That makes the problem hard to find. ioctl analysis told me nothing, no relevant ioctl error found. I guess the kernel driver could be not 100% up-to-date, so it misses some bits, some Ivybridge specific stuffs, but it's just a feeling. It's a hard topic. Everything is well recognized (you have also to force AIGLX activation, it's disabled by default with a patch in xorg-server), chip and mesa library and so on, BUT: if you test some mesa-demos, you could recognize that the swrast is used (engine, for instance, shows typical artifacts of the software raster), instead of the right one. With Sandybridge, no problem. Best regards, Luca