On Wednesday 24 Sep 2003 4:39 am, Lance Cummings wrote: > First, a quick thank you to everyone that's chipped in on this. > > I've run across, and also been pointed toward, several modeline > generators. I present to you three of them: > > 1. http://www.sh.nu/nvidia/gtf.php > > 2. http://www.dkfz-heidelberg.de/spec/linux/modeline/modeline2.cgi > > 3. http://xtiming.sourceforge.net/cgi-bin/xtiming.pl > > All of them offer to pump out a modeline for you after you input the > values of the Fh and Fv ranges, and the resolution and refresh you > want to run at. Wow, what convenience. > > Using the values of Fh = 30-97, Fv = 50-160, and 1024x768 and 100 Hz > refresh however (you know what is coming, right? <vbg>) produces the > following three DIFFERENT modeline calculations: (numbers correspond > to the URLs above) > > 1. # 1024x768 @ 100.00 Hz (GTF) hsync: 81.40 kHz; pclk: 113.31 MHz > Modeline "1024x768_100.00" 113.31 1024 1096 1208 1392 768 769 > 772 814 -HSync +Vsync > > 2. # V-freq: 100.00 Hz // h-freq: 81.70 KHz > Modeline "1024x768" 124.84 1024 1088 1240 1528 768 768 771 817 > > 3. Modeline "[EMAIL PROTECTED]" 126.64 1024 1056 1536 1568 768 781 794 807 > > I ask you . . . are we having fun yet? The values wildly disagree > with each other, but since this is presumably a precise mathematical > calculation, only one of them can possibly be correct, but indeed > NONE of them NEED be correct of course. It's possible they are all > wrong.
Not at all. They are all using different dot clock frequencies to give you the same display. > > Certainly someone in "the community" who is really "in the know" > ought to have a look at these discrepancies, I would think, and set > the record straight for everyone else? They all look reasonable to me. I dont count myself a video expert, but I do understand this much and what the numbers mean. I cannot find anywhere in the thread a mention of which video card you're using. What is its maximum clock rate? The Viewsonic website actually quotes both of the frequency ranges. the 90kHz/180Hz range on the HTML spec, and the 97kHz/160Hz on the PDF spec. Since both specify a bandwidth of 200MHz that seems safe. I don't think you need it though, it's the bandwidth of the card that you need. The output of a "startx --probeonly" would be very helpful. It would tell us the rates that the monitor itself was happy with (if for example viewsonic changed the spec halfway through production.) It would tell us the dot rates that the card was capable of, and it might even tell us the specifications of the sync pulses. > I certainly don't want to plug the wrong modeline in and fry a very > nice monitor. And I'm simply not qualified to calculate my own > modeline, since I don't clearly understand the math behind it. > Obviously, from the results above, I'm not alone on that score, and > even people who think they know what they are doing well enough to > put up programs for others to use may indeed be flailing around in > the dark. Unless I am completely mistaken about this concept (a > possibility, I readily admit) at least TWO, if not all three, of the > modelines generated above are worthless, and maybe even dangerous to > use. No. They *could* all be right, and I don't see anything to suggest that they aren't. They all choose a different dot clock frequency (113.31MHz, 124.84MHz and 126.64MHz) They could all choose a different length for the horizontal and vertical sync pulses, (I haven't done the maths.) Depending on your monitor and your video card one might be better than the others, or they all might produce the same visual effect. If you were you, I'd try all three. If one works then we don't need to bother going any further. If not then send me the output of "startx --probeonly" and I'll extract the interesting bits for the list (IIRC it's rather long.) And also post the spec of the video card. -- Richard Urwin
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
