Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!news.ecn.uoknor.edu!feed1.news.erols.com!super.zippo.com!zdc!zippo!drn From: tedm@agora.rdrop.com (Ted Mittelstaedt) Newsgroups: comp.unix.bsd.freebsd.misc Subject: Re: Problems: EISA+1542; NE2000(PCI); onboard MACH-64 Date: 8 Jan 1997 14:00:06 -0800 Organization: Zip News Lines: 85 Sender: usenet@drn.zippo.com Message-ID: <5b15d6$1sn@lana.zippo.com> References: <32CE572F.2781@datatone.co.uk> NNTP-Posting-Host: 205.184.255.74 Xref: euryale.cc.adfa.oz.au comp.unix.bsd.freebsd.misc:33864 In article <32CE572F.2781@datatone.co.uk>, Dave says... > >I am having several problems with installing FreeBSD on two separate >machines. > > >This is a Pentium PCI based machine (Intel motherboard, I believe >marketed as 'Atlantis'). The motherboard has an on-board ATI MACH-64 >graphics chipset with 1Mbyte display RAM. I have a PCI NE2000 clone >network card. > >2) I have not managed to get the X-server to operate as a MACH-64. > >The startx output (saved to a file) indicates that the Mach-64 server >cannot find a dot-clock to match either 1024 or 800 resolutions, but >settles on 640x480 resolution. There appear to be no other obvious >errors and the last line in the log file says: > "Apperture mapped to 0x4000000" >The only other thing that might be significant is that I did not enter a >RAMDAC configuration parameter and I get a "Ramdac is Unknown (0)" >message followed by "Using 6 bits per RGB value". Is this why I get no >display? Do upgrade to the latest release (2.1.6) and the X in it, as Joerg suggested. However, a couple things I learned about Xfree: 1) The chip probe routines suck. What you want to do is carefully read the various README.XXX server files in /usr/X11R6/lib/X11/Server/doc that pertain to the chipset that you think that you have, then you open the machine up and identify the video chips. Your going to want to find a total of three chips, the RAMDAC, the Video chipset, and the Clock chip. On ordinary VGA, or el-cheapo clone VGA cards that use fixed clocks you might only be able to locate the video chip, but on higher-quality super VGA chipsets you should have all three. 2) The different X servers included with Xfree are of varying qualities. For example, the S3 server is quite complicated, with support for many options, but certain chip combinations have to be present for some of the options to work. On the other hand, the Oak server is rather simple, with support for only a few options. It is not uncommon for the various server probes to be unable to reveal the appropriate ramdac configuration line, and if this is missing most of the accellerated servers will upchuck if they cannot autoprobe the ramdac. S3 seems to be the worst of the lot, it's readme file contains dozens of ramdac IC numbers to look for on your card, and the server cannot probe half of them. In many cases, the individual chipset servers don't support all the options present in the general X documentation. Use the readme's for the servers as your final authority. 3) The monolithic SVGA server is non-accellerated, but it has support for many different chipsets all crammed in to the server. In contrast, the accellerated servers are pretty much single chipset servers. You may have better luck getting X running by starting out with the SVGA server. Obviously, it won't be accellerated, but you can often use the generated XF86Config file as a base to get an accellerated server operating. 4) The old version of X used the "xf86config" command which either used the SVGA server or a user-selected chipser server, then ran a probe with that server for various XF86Config file options such as clocks lines. The new version of X has this graphical setup program, xf86Setup, which attempts to probe and determine the desired chipset by itself. The new X version also retains the older xf86config program. In my case to get an operating server I had to run both setup programs, neither of which properly probed the chipset, and hack and slash the resulting XF86Config files together to get an operating server. One gotcha is that the new program will default to the SVGA server if it's inital probe cannot figure out the chipset, and if you don't install that the setup program will bitch and complain, so make sure that server is installed. One nice thing I'll say about the new program is it created a far cleaner XF86Config file, the older install program's XF86Config file was a mess and very difficult to understand exactly how it operated. Unfortunately, the win95 video detection is only good for telling you the chipset, or what it's attempting to clone, that's only part of what is needed. Video card manufacturers all have a vested interest in keeping video as difficult as possible to deal with, and as incompatible with each other as possible, if video ever were standardized the card vendors would all go broke.