Kernal Panic Attacks - how to diagnose

Question

I have recently started to get kernal panics on my dual processor G4 with increasing frequency. The problem started when I switched monitors to a 1600x1200 Samsung (analog connected). They went from once every few days, to every day, to multiple times a day, to the point where my HD could not complete a start up without a kernal panic during the apple screen.
I tried the following:
1) tried the Apple Hardware Test CD and it check out fine.
2) Ran Repair Disk from OS X Tiger install disk - no errors found
3) Ran Repair permissions - many errors found and corrected
4) Ran repair permissions again - no errors found
5) Panic occured while still booted from Install disk
6) Machine seemed to work when started from external HD (though only used for a short period)
7) tried to boot from internal HD again and got panic before apple screen complete
8) removed memory (3 - 512K) and tried with individual dimms only. No panic each time - startup OK.
9) restored all memory and system now seems to work
Not sure if I will get panic again, but at least I can boot from internal HD again. Any ideas as to what else I can check. I believe I have gone through Dr Smoke's steps. Could problem have been due to dust on memory?
Panic log:
Thu Sep 14 20:42:19 2006
panic(cpu 1 caller 0x000A8B80): Uncorrectable machine check: pc = 00000000000A4D80, msr = 0000000040001030, dsisr = 42000000, dar = 00000000000BF000
AsyncSrc = 0000000000000000, CoreFIR = 0000000000000000
L2FIR = 0000000000000000, BusFir = 00000000ffffffff
Latest stack backtrace for cpu 1:
Backtrace:
0x00095564 0x00095A7C 0x00026838 0x000A8B80 0x000A7E10 0x000AB880
Proceeding back via exception chain:
Exception state (sv=0x46D0E780)
PC=0x000A4D80; MSR=0x40001030; DAR=0x000BF000; DSISR=0x42000000; LR=0x00032BB0; R1=0x22103DB0; XCP=0x00000008 (0x200 - Machine check)
Backtrace:
0x00032B78 0x0003C7DC 0x000A9654
Exception state (sv=0x42566780)
PC=0x00000000; MSR=0x0000D030; DAR=0x00000000; DSISR=0x00000000; LR=0x00000000; R1=0x00000000; XCP=0x00000000 (Unknown)
Kernel version:
Darwin Kernel Version 8.0.0: Sat Mar 26 14:15:22 PST 2005; root:xnu-792.obj~1/RELEASE_PPC

Answer 1

The "uncorrectable machine check" and Machine check exception state is pretty much an unrecoverable hardware error indicative of a more serious issue. It could be bad memory, it could be the logic board, it could be a processor issue, combination of some/all. Normally something like this may occur when the system is overstressed but faulty hardware exasperates the problem under even lesser loads.
In my case, I had intermittent panics which were hard to track down (pretty much folllowing the steps you did). I eventually did rule out the ram by running them in my Cube without any problems. Sometimes I could go weeks without an issue even under load after doing the troubleshooting but then all of a sudden, every few minutes to a few hours would result in a KP doing something simple as web browsing. Eventually it got so severe that one day, I got that machine check panic as soon as I powered on. It turned out to be one of the processors on the dual processor board failing from overheating.
This happened twice on my QS2002 within the first two years. The first time, the system could be coaxed to boot after letting it sit since the failing processor wasn't completely fried yet but the second time (not too long after the initial incident), one processor burned out completely (machine check KP right at power on). The tech showed me the chip and how bad it was. Luckily the first two times were covered under Applecare.
The third time happened out of warranty (at which point I was ready to grab the machine by the handle and fling it out a window out of frustration). Was almost ready to pull the trigger on a PM Dual G5 but decided to get the cheapest dual G4 (1.2GHz) upgrade from GigaDesign at the time in 2005 since this would hold me over until the Intel Mac's came out.
In all three cases, heat (even with attempted improvements to the airflow/exhaust) was the primary factor in the eventual processor failures since my system was always under constant load. I think the culprit was insufficient cooling of the aluminum heatsink (the GigaDesign uses a better designed and smaller copper heatsink along with two attached fans) which is magnified by a case which already has poor airflow.
I've had no problems with the machine since this brain transplant. Of course, anyone who is able to narrow it down to either the processor or logic board needs to think if it is worth sinking any more money into at this juncture considering the prices of a new Mac.
Mac OS X (10.4.7)