On Mon, 8 Mar 2004, William D. Hamblen wrote: >I promised I would write back with the results of my attempt to assemble a >RHEL3 system with two flat panels, both using DVI connections. So here it >is. > >The short version: two Radeon 7000s (one AGP, one PCI), two Dell 1703FP >1280x1024 flat panels. Works great out of the box if you disable DRI in >XF86Config. No artifacts, stable image, uniform color across both >displays. Very nice. > >I ended up ordering a prebuilt system from a vendor we've used fairly >often. It came with two 64MB Powercolor Radeon 7000 Multi-Display Edition >video cards (Evil Wizard): one AGP and one PCI. The vendor picked the >specific card - all I cared about is that they were Radeon 7000s with DVI >(because they are cheap and have a PCI version). These cards also have a >VGA connector and TV Out. lspci calls them > >01:07.0 VGA compatible controller: ATI Technologies Inc Radeon RV100 QY >[Radeon 7000/VE] >02:00.0 VGA compatible controller: ATI Technologies Inc Radeon RV100 QY >[Radeon 7000/VE] > >I added two Dell 1703FP flat panels. These have DVI connectors and a >native resolution of 1280x1024. > >I did the Redhat Enterprise Linux WS 3.0 installation myself. There was >only one problem. After the installation finished, the system rebooted >and the primary screen (only one is setup by the installer) went blank as >X started. The system appeared to be completely hung. Nothing on any of >the four video connectors (2 DVI, 2 VGA) and I couldn't switch to a >virtual console (even once I bypassed the firstboot script and started >"for real"). Yet, the monitor connected to the primary card's DVI output >was reporting a 1280x1024, 60Hz signal. > >I should interrupt myself here to mention that the system came with an >nForce2 based motherboard (ugh, my bad for not specifying) so I had no >ethernet support to login remotely and see if the system was really alive. >Unfortunately I did not think to save the XFree86 logs - but they looked >normal to me. > >I did the RHEL WS installation several times (out of curiousity - at this >point I knew what was wrong) to see if the installer would come up with >something that worked with DRI. The first install was with both cards in >place and the BIOS set to use the PCI card as primary. The second try was >with just the AGP card (BIOS still set to PCI since I hadn't seen the >option yet :-)). The third try was with just the AGP card and BIOS set to >AGP. > >Those all hung in the same way after the initial reboot. I did a fourth >installation with both cards installed and the BIOS still set to AGP. But >right after the installation, and before the reboot, I switched to a >virtual console and disabled DRI in XF86Config. It worked beautifully >from there. > >I added the usual stuff to XF86Config and both screens look great, with >DVI, but no DRI (which I do not care about in this application). I can >send the XF86Config file to anyone who cares. I merely turned on >Xinerama, added the second Device and Screen sections and inserted BusID >lines. > >I hope this is helpful to someone. Just as a point of reference, DRI is disabled if you are using multihead because DRI is not multihead compatible. So all multihead setups have DRI disabled wether or not Xinerama is being used, and also wether or not both heads are on one card or several. It is functionality that is just not implemented in XFree86 currently. Having said that, several drivers have had "merged framebuffer" support added to them in upstream development, and in a future OS release, mergedfb enabled drivers will support multihead+xinerama as well. Just thought I'd mention that so nobody wastes time trying to make DRI work with Xinerama. ;o) _______________________________________________ xfree86-list mailing list xfree86-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/xfree86-list IRC: #xfree86 on irc.redhat.com
