Since last posting, I discovered that the window manager is not at fault; the display failure occurs also when using blackbox.

I followed opjose's advice about downloading and installing ati-kernel-2.6.8.1-12mdk-3.14.1-1mdk, but I had to revert to using the 2.6.8.1-12mdk kernel instead of the 2.6.8.1-20mdk one I was using, as there was no version of the ati-kernel RPM for the later kernel version. I don't know if this means the ati
drivers in this RPM are built into the later kernel, or if they just haven't been able to provide the same modules for the later kernel version yet.

Yesterday I increased the AGP aperture size to 128 Mb from the default 16 Mb, after reading about it and (wonder of wonders) actually understanding the issue involved. Graphics performance does seem a little snappier since that move. I should note again that I don't do any heavy graphics-intensive stuff like huge games; just image editing, word processing, spreadsheets, browsing, the usual mundane non-game activity of a computer potato.

I also moved my ethernet card a couple of notches away from the graphics adapter card; they were not side by side (which I've heard is definitely not the way to have them) but there was only one empty slot in between them.

After a while the system locked up and the monitor wouldn't even TURN ON. I replaced it with the 15-inch monitor I was using the last two years, and it worked OK. I then moved the ethernet card one slot up--now there are two empty slots between it and the graphics card. Put the 19-inch monitor back in place and so far so good.

I do remember that when an SBC technician was out here troubleshooting DSL failure once (turned out my modem was fried by lightning, and also the ethernet card I was using at that time!), he found that one slot didn't want to work with the new ethernet card. Maybe that was the one I tried this morning that caused the lockup.

I'll see if these ATI modules remedy the problem, though I have nothing like high hopes....

Eh, it could well be that more than the modem was affected.

Your system may also experienced enough degredation to affect say AGP video transfers so it's resorting to PCI...

I've seen this happen with lightning strikes.

The lightning episode happened well over a year ago and I never had the screen-garbage/blackout/slow repaint problem till I installed Mdk 10.1 about three weeks ago... Still of course you could be right and it could be a coincidence, with damage just now manifesting itself visually.

In a classic example of locking the barn door after the horse (or cow, or whatever) escapes, I installed a TrippLite line-conditioner/surge/sag protector within a week after having my modem and ethernet card replaced. Then more recently I installed an APC battery backup unit--and a week after that, a storm knocked out power while the computer was running, and everything got gracefully shut down! Two very good investments.

I'm beginning to think this display problem is a tactic to get me to buy a new computer.

I thought maybe the ati-specific kernel addon I downloaded and presumably got installed (how would I know?!) fixed the problem, but no, there was another glitch: crazy pattern, brief blackout, back to normal, all in maybe ten seconds. So that didn't do it.

I ran a memory test for ten hours and a half and got not one single error, so the RAM seems OK.

Next step, I guess: hook up the 19-inch monitor to my old decrepit Windows (shudder) machine and see if it behaves; use the tried-and-proven 15-incher it replaced with this Linux setup and see if it FAILS--in which case the new monitor is probably not at fault, but software (which I'm fairly certain is the case).

This needs patience on a level somewhere between that required for breeding pandas in captivity, and reading a Stephen King novel--with the latter in top place. (OK, I admit I'm not a SK fan.)

If it's a new DRIVER I need, I'm probably just out of luck at least till maybe Mdk 10.2....if ever.... I would think there are enough ATI Rage 128 Ultra TF users around that the card would be supported pretty well, though. Maybe it IS supported pretty well.

The "glitch" pattern (instant of blackout, patchwork color, brief blackout, return to normal) has not occurred since I noticed something about three days ago.

I am phonophobic (phobia of noise) and consequently wear earplugs almost all the time. I was sitting at the computer while not wearing earplugs and the glitch occurred. Simultaneously I heard one of my three hard drives spin up to operating speed.

It seemed to me more than a coincidence. Why X would need a hard drive I have no idea--the swap drive is almost never used, and was not used in this case; I have 384 Mb of RAM and only run modest applications that usually leave at least 1/3, usually more than 1/2, of the RAM free.

So it seemed to me that, whatever the reason for X needing the hard drive--if that was the case--it was unavailable because of power management having spun it down.

I went into the BIOS and disabled power management. I found out that Linux couldn't shut off the machine at the "poweroff" command that way, though. So I changed power management settings to high values, and while I was at it, I changed all the IRQ settings to "monitor" in case something was getting missed that was happening.

No problem with this since. Not one time has there been an interruption.

However, at powerdown the background still sometimes (not always) changes from Mandrake Blue to the crazy-quilt appearance of "random" pixels. So I'm not convinced the problem's all the way solved.

I also don't know why under Mdk 10.1 X seems to operate differently from under 9.2, where I had no problem. Maybe it's the switch to Xorg?