Return to BSD News archive
Newsgroups: comp.os.386bsd.questions Path: sserve!manuel.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!pacbell.com!att-out!cbnewsj!dwex From: dwex@cbnewsj.cb.att.com (david.e.wexelblat) Subject: Re: How to select 72MHz on ET4000 board? Organization: AT&T Date: Fri, 5 Mar 1993 17:21:57 GMT Message-ID: <1993Mar5.172157.6332@cbnewsj.cb.att.com> References: <1993Feb25.151244.15114@gmd.de> <rcpt.730720661@rwa.urc.tue.nl> <1993Mar4.171002.5808@super.org> Lines: 77 In article <1993Mar4.171002.5808@super.org> becker@super.org (Donald J. Becker) writes: > In article <rcpt.730720661@rwa.urc.tue.nl> rcpt@urc.tue.nl writes: > >veit@borneo.gmd.de (Holger Veit) writes: > >>In article <rcpt.730581552@rwb.urc.tue.nl>, rcpt@rwb.urc.tue.nl (Piet Tutelaers) writes: > >>|> > >>|> The ET4000 board in our 486DX33 box does provide a 72MHz clock in > >>|> order to achieve a 70 Hz refresh rate for 1024x768 mode. The manual > >>|> explains how to select this frequency with VMODE.EXE an MSDOS utility > >>|> (mode 0x38). > >>|> > >>|> If I startup `startx' and write its output into a file I don't see this > >>|> 72MHz clock. Does this mean I can't select this clock with XFree1.2? > > > >>The clock might be select with a mechanism that is not in the area > >>of the standard 16 clocks. This is not supported by XFree86. Another > >>possibility is that your VMODE (or the handbook) lies about the 72MHz; > >>also already seen. There is a facility to run an external clock > >>program (if you manage to find out how the 72MHz clock is enabled), > >>see the XFree86-1.2 doc on this. > > Xfree86 doesn't usually detect the correct clocks in my ET4000. It > has some clocks that are far abover the 65Mhz-nominal-max rating of > the ET4000, but it Xfree often screws up the lower-rate clocks as > well. You can either > > o run Xfree a bunch of times on an idle machine, and guess at which > values are correct, or > o Add a 'Clocks' line that just numbers the clocks 0 to 15, and add > 'modes' and resolution lines that use every clock. Step through the > clock rates to find one that you like. > > Clocks 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 > Modes "1024x768-0" "1024x768-1" "1024x768-2" "1024x768-3" "1024x768-4"... > > If you are curious about the true dot-clock rates, measure the > horizontal sync with a frequncy counter and multiply away... > > DB > -- > Donald Becker becker@super.org > Supercomputing Research Center > 17100 Science Drive, Bowie MD 21114 301-805-7482 XFree86 1.2 has improved the accuracy of the clock probing. Have you tried again with 1.2? Many boards have 80MHz clocks. I think that is is part of the problems that some users see with snow/smearing on XFree86. The spec for the ET4000 is 65MHz, but the board is supplying higher clocks. The reason that you see the REALLY high numbers (100+Mhz) is the divide-by-2 bit. These clocks really do exist, but the BIOS doesn't use them (nor should you). The BIOS sets the ET4000's divide-by-2 bit (hence getting a 65Mhz clock from a supplied 130MHz clock). This makes it much easier to generate clocks from 25Mhz to 80MHz - just use the divide-by-two when you need to. XFree86 detects 16 clocks for an ET4000, even though it only has 3 clock select leads. The reason is that we use the divide-by-two bit as an "extra" clock select. The other thing that I just discovered on the WD chipsets, and will investigate for Tseng as well, is that some chipsets allow you to feed MCLK as VCLK (i.e. use the memory-refresh as the video dot-clock). Which means that there are really more clocks available than XFree86 is detecting. I will be looking at this for all of the chipset for which I have documentation (not all of them, unfortunately), and adding support for this feature to XFree86 1.2.1. Anyhow, this may explain where the 72MHz clock is coming from. -- David Wexelblat <dwex@mtgzfs3.att.com> (908) 957-5871 AT&T Bell Laboratories, 200 Laurel Ave - 3F-428, Middletown, NJ 07748 "In and around the lake, mountains come out of the sky. They stand there." -- Yes, Roundabout