Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!sdd.hp.com!spool.mu.edu!snorkelwacker.mit.edu!stanford.edu!leland.Stanford.EDU!fangchin From: fangchin@leland.Stanford.EDU (Chin Fang) Newsgroups: comp.unix.sysv386 Subject: How to config your vedio subsystem for X386 (was Re: X386 1.1) Summary: I hope this tutorial turns out to be helpful Keywords: X, Roell, X386 Message-ID: <1991Mar7.014948.18583@leland.Stanford.EDU> Date: 7 Mar 91 01:49:48 GMT References: <1991Mar6.213907.28719@maverick.ksu.ksu.edu> Organization: AIR, Stanford University Lines: 183 In article <1991Mar6.213907.28719@maverick.ksu.ksu.edu> rrsum@matt.ksu.ksu.edu (Rick Summerhill) writes: >I am currently using Tom Roell's X386 1.1 with ISC 2.2.1 and have some >questions: > >1. I get errors from Xlib about too many clients. I have 32 pty's and > 32 streams ports, so that is not the problem. Anyone know why? I don't have ISC, can't help you on that. But... >2. Does anyone have a good 1024x768 configuration for a Sigma Legend card. > My monitor will go to 66 Mhz. Can 1152x900 be used in non-interlaced > mode with this monitor configuration? OK. I think it's time to sweep away some confusing terminologies introduced by marketing people at monitor manufacturers. Monitors NEVER support Mhz which is by definition, million cycles by second. Your monitor, if multi-scan- ing, will support khz in it's horizontal sync frequnencies, which by definition is kilo (or thousand) cycle per second. In the vertical direction, the sync frequency is always in the order of 10 hz (eg, 15 hz ~ 75 hz, like NEC 4D). However, if analog type like NEC 2A, only descrete frequencies are supported. Not a continuous range. So be careful! NOTHING in the world of monitors will support MILLION hertz range support sync frequency! It's your vedio adapter which MAY support such a driving frequency! (see below) Furthermore, it's up to your software driver to see if such a driving frequency can be utilized or not! When we talk about display, it's always NECESSARY to consider three things together: (1) your monitor's sync frequency ranges for both horizontal and vertical (2) your vedio adapter's driving frequency bandwidth (from crystals for ex.) (3) your software's vedio hardware device driver. Anyway, I regard this frequently advertised usuage as bad as, or even worse than using Mhz to denote data transfer rate in hard disk ads. Hz, named after the German who did quite a lot ground work in EE theory, SHOULD be used to describe oscillatory phenomenon with fixed frequencies. Any other usuage is INCORRECT! How to determine a good resolution for your monitor? Please read the following Definitions 1. screen refresh rate => it's the DRIVING clock frequency divided by the product of horizontal frame length and vertical frame length. Elaborations: Q. what is driving clock frequency? A. it's the oscillatory frequency of the crystal(on your vedio board) employed by your graphics software's vedio driver. For instance, if your driver uses 36 Mhz out of 25 28 0 40 36 40 45 58 32 36 31 35 50 48 33 65 (all in Mhz, for Sigma Legend) the the driving frequency is 36Mhz. Nothing else. This is the frequency used by the driver to determine how frequent to update (thus refresh) screen image. Q. What's frame length for horizontal and vertical directions? A. It's the number of clock tics (measured in terms of the driving timing) for your monitor's electron gun to shoot a beam of electrons from left side of the tube to the right side and back. Similarly, for vertical direction, from bottom to top and back to bottom. Q. Why this has anything to do with resolution? A. because it's basics in highschool physics. So you MUST understand it! (to answer your son's questions). Your desired resolution in fact should be called the portion of frame length that electron beam creates visible image to your eyes! Any image is created with your eyes's retention and the fast moving electron beam sweeping thru your monitor. At any instance, there is in fact ONE dot hitting the screen, but due to your eyes retention, you see a block of image. Q. So what does this have anything to do with screen refresh rate? A. By definition, one hertz (hz) is one cycle per second. So, if your horizontal frame length takes x tics, ver. frame length takes y tics, then to sweep thru the entire screen, a rectangular area, takes x times y tics. Since your driving frequency provides say N tics per second by definition, then obviously your monitor's electron gun(s) can shoot a dot accross the screen and sweep it from left to right and back and from bottom to top and back, which takes total xy tics, N/xy times. This IS your screen's refresh rate! Because that's how many times your screen can be updated thus REFRESHED per second! Q. Why I have to know this B.S.? A. Because anytime you use a refresh rate lower than 60 Hz (in layman's words, sixty times every second), you are trying to demage your eyes! So that you won't be enjoy gif pictures as you may like in the future!! ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Q. So how do I get a desired resolution? A. Simple! Jut look at your monitor's data sheet, normally part of your users' manual. Make sure it's type, ie. analoy or multiscaning. The later is much flexible if not better. THE FOLLOWING DISCUSSION DOES NOT APPLY TO ANALOG TYPE MONITOR!! Then, use your horizontal max sync frequency, say 55khz try out the clock frequencies listed or detected by clock.exe. As an example, say for Sigma Legend, there is a 65 Mhz clock oscillatory driving frequency available. And from above assumption, your monitor can sync up to 55 Khz in horizontal. To increase max REFRESH rate and at the same time get non-interlaceness, do the following Obviously, if your clock cycles only N times per second, and if your monitor electron beam syncs at x times per second, the using the definition of frame length above, you can only have so many horizontal frames per second: N/x, in our sample, it would be 65Mhz/55khz=1181 times. But due to VGA's hardware restrictions, you can only have multiple of eight number of frames per second in the horizontal direction. So round it off -> 1176. Take 80 percent of this clock tics, or 944 tics for your viewing image. This is a rule of thumb! Don't ask why! Take 75% of 944 as your vertical tics for viewable image, you have 708 tics. A rule of thumb is 1.05 times of tics should be the vertical frame length ->743 tics. Here I implicitly assumed that you like the Golden ratio. So your screen refresh rate is 65Mhz/1176*743=74.4 hz! THIS IS EXCELLENT! Don't settle on anthing less! The image at this update rate (or screen refresh rate) WILL LOOK STEADY EVEN TO VEDIO CAMERAS! And you got 944x708 to boot. Not bad at all! You can even improve it further to put it into almost 80 Hz by using the fact that your monitor in horz. direction can sync at lower frequency then 55 khz. Please take advantage of it. All this is simple arithematics and Simple facts about oscilloscopes. No black magic at all! Memory requirment: 944x708/1024~653 K vedio RAM. So if you have one meg, you have extra for virtual terminal switching. See, this is good compromise! But MAKE SURE that your monitor electron guns can sync up to 80 Hz vertical. (NEC 4d CANN'T for instance. It goes only up to 75 Hz in vertical) There is NO reason whatsover why you have to use 640x480 800x600, or even 1024x768. X386 driver lets you config your hardware with quite a lot freedom. It usually takes two to three minutes to come up the right one. The KEY is high refresh rate with reasonable viewing area. NOT Hig Res at the price of flickerness! Q. But how about interlace/non-interlace? A. The key word in vedio watching is NON-FLICKER! The point is that non-interlace is just part of the game! With non interlace alone BUT low screen refresh rate, your eyes will suffer badly!!!! Interlace-ness just worsen the flickering at the same refresh rate. If you can manage to get high enough refresh rate, say 90 Hz for interlace display, you WILL NOT feel any flicker!! (But I doubt this is feasible) As a side, that's why I HATE the display of IBM/RS6000 even with it's blazingly fast TI340X0 assisted graphics! So, CONCLUSION => NON-INTERLACENESS != NON-FLICKER! [question deleted, I don't have ISC] Please direct your comments to the above article to the addressee below: Chin Fang Mechanical Engineering Department Stanford University fangchin@leland.stanford.edu