Horun wrote on 2021-07-07, 00:05:

Have never seen that exact error from either Win98 or WinXP.
Which OS gave you that error ?

I've seen that kind of BSOD on Windows XP during my school days when a computer was attempting to shut down. Never saw that type of BSOD ever again until now.

antillies wrote on 2021-08-15, 17:20:

I guess I'm curious what caused the original to fail. I suppose it could have just been at the end of its life, but I also wonder, since I had been running my 6200 prior to the issues, if the heat from the GPU caused or accelerated the card's decline? If anyone would have a postulation, wouldn't mind hearing any.

I believe that it could be related to lifespan like you say due to many years of continuous use prior.

Bad news.

I was playing Battlefront this evening and everything was running just fine until...it froze. Same screech from the speakers as the first time (and many times since).

When I manually powered off and powered back on, my monitor lost the DVI source while booting. At this point I was thinking I had guessed wrong and it was, in fact, the GPU, so I swapped my 6600 GT out for my 6200.

Source returned and things seemed like they were going smoothly until it froze again, to this screen:

Tried restarting but again lost source mid-boot.

I am flummoxed. So far I've changed out the motherboard, the GPU's, and the sound cards. The RAM tests fine. I suppose I could be unlucky and have multiple bad duplicates, but that seems unlikely to me. Once things cool down I'll run a stress test on my 6800 to see if any issues crop up there.

Could it possibly be something to do with having two OS' on the hard drive? Would it be worth wiping and installing only XP?

I hate it when things like this happen. You could very well just have multiple problems and it wasn't just 1 thing. Sounds like the old and new GPU are both bad.

These problems are really hard to solve. I have had this problem before too twice. Once it was broken cable (ide) and once a CPU. (those P4 actually can somehow cook themselves to death)
However, seeing the balloon pop up that there is new hardware found makes me suspect your VGA/Motherboard/PSU line.
Could even be as simple as a bad contact in the grapics card slot, bending under weight of card/slot, worsened by vibrations.
But you really have to work your way back to the minimum to test this. So Motherboard/CPU/RAM/VGA and then check what happens.
Plus if you can test it outside it's case, then you can also move/wiggle to see if this causes it.

I hope you can pinpoint it and update us on a stable system again.

I'm leaning towards both GPUs being bad. I recently bought 2x GeForce 6600GT cheap, thinking at least one will be good. First one doesn't even boot, the other one displays artifacts. It's quite normal with these high performance cards. It's a hit or miss which is why I never spend more than €8 on them. If you shell out on an expensive one, you may be disappointed with random crashes like these as nobody tests these cards properly.

Thanks for the responses.

I think I'm going to try replacing the CPU (since it's the only piece of hardware I have yet to swap out) and see if that fixes it. I will report back once I've done that.

IMO 4 sticks of DDR in any board is always a rodeo. They will have a "subtle" problem in that it won't be a constant bit error, it will be all over the place according to what operations happened on that stick and a neighboring one in the recent past, memtest doesn't test for this kind of stuff until test 7 or 8 so passing test 1 doesn't tell you anything. It's likely not the fault of the RAM per se, just ain't 100% on that board/chipset. It can be patiently tuned out by relentless reboots to memtest, running the tests that are getting errors, tweaking the settings in CMOS setup and going back and trying again. So that's a whole day worth of messing around, or just cut down to 2 sticks. I'd almost place money on either of those pairs running fine just as a single pair.

BitWrangler wrote on 2021-08-19, 18:56:

IMO 4 sticks of DDR in any board is always a rodeo. They will have a "subtle" problem in that it won't be a constant bit error, it will be all over the place according to what operations happened on that stick and a neighboring one in the recent past, memtest doesn't test for this kind of stuff until test 7 or 8 so passing test 1 doesn't tell you anything. It's likely not the fault of the RAM per se, just ain't 100% on that board/chipset. It can be patiently tuned out by relentless reboots to memtest, running the tests that are getting errors, tweaking the settings in CMOS setup and going back and trying again. So that's a whole day worth of messing around, or just cut down to 2 sticks. I'd almost place money on either of those pairs running fine just as a single pair.

I appreciate your input. Others have also suggested maybe the large amount of RAM is the culprit behind this (particularly because I'm dual-booting 98 and XP).

I did test it though two nights ago when I was again experiencing problems by removing all but one of the sticks. Turned the computer back on but it again froze on boot-up, so I don't think it's that (though I do wish it was something simple as that).

If you have changed all parts and the PC keeps failing the same way, I'd bet the issue lies in one of the parts you have not changed. It may sound silly, but try a different power supply and different HDD cables. Yes, cables can go bad. Also, I've seen memory that passes memtes86 but fail when used on common tasks, no clue why it happens but it does.

Another month and another update.

I did end up purchasing another CPU. I paired it with my new motherboard but ended up removing both form the build after a bit of a misadventure that ended in something smelling like smoke. No idea if I accidentally fried the CPU, the motherboard, or both, but decided it wasn't worth continuing tests with that particular combo for fear of potentially damaging more components.

Misadventure aside, I ran into a strange issue off the bat where after I installed the CPU I wasn't getting signal to the display. I ended up putting the other CPU back in and testing but got the same result. I reset CMOS but again no luck. I ended up reseating the RAM and that did the trick. I put the new CPU back in and it seemed to do okay at first but when I loaded up Battlefront I again lost signal. I left it for the night and when I reapproached the next day I was getting inconsistent behavior.

At this point I decided that the benefits of maintaining the Windows 98-XP dual environment was no longer worth the cost of dealing with all these issues. I decided I needed to start fresh to see if the extra memory and accompanying workarounds were a potential cause. I was actually in the middle of reinstalling XP when the system crashed and I smelled smoke.

With my original motherboard and CPU back in the case, I proceeded with a fresh install of XP with only one stick of RAM. XP installed without issue. I loaded up the drivers (much easier this time without having to negotiate two separate operating systems) and installed Battlefront again. Played for about forty minutes and, aside from Battlefront itself crashing, the system performed just fine. I tested it again this morning without issue.

I think it's too early to say whether I am out of the woods completely on this but I'm hopeful given the initial positive results. Since all the hardware testing seemed to show no glaring issues, I am persuaded that the problem ultimately sat with the the dual-boot environment I set up and the workarounds I implemented. I suppose it's also possible the system did work initially but became unstable over time with use.

If the build continues to work well with the fresh install, I think that will point to the above being the case. Although I am somewhat sad to no longer have 98, I feel better knowing that things are (currently) more stable. If I ever feel the need for an 98 machine again (and have the space), I'll make a separate machine. Glad to have accomplished the dual-boot at least once but I don't think I need to attempt it again.

Thanks everyone for their continued help and comments. I'll report back after a bit and let everyone know how things are going.

In my experience many weird errors can be traced down to power supplies. If you dont have clean power you will rarely have a stable system.

I know so many people that put tons of $$ into a system then put in the cheapest power supply they can find and then wonder why their expensive systems is so flaky.

Grem Five wrote on 2021-09-21, 20:51:

In my experience many weird errors can be traced down to power supplies. If you dont have clean power you will rarely have a stable system.

I know so many people that put tons of $$ into a system then put in the cheapest power supply they can find and then wonder why their expensive systems is so flaky.

I agree with you. The PSU is one of the most important part of the computer that sometimes gets neglected, even on cheaper desktop computers one must not skimp on this for stability reasons.

One has to stay away from generic PSUs, all of them have fake specs and cut corners on components.

Grem Five wrote on 2021-09-21, 20:51:

In my experience many weird errors can be traced down to power supplies. If you dont have clean power you will rarely have a stable system.

I know so many people that put tons of $$ into a system then put in the cheapest power supply they can find and then wonder why their expensive systems is so flaky.

I bought a Corsair CX450M 450W 80 Plus Bronze for this machine. Not the best, but also not the worst.

Would you have a recommended way to test whether the power supply is causing issues, besides replacing it? I'm certainly not opposed to upgrading it if the system continues to act up. As of right now, though, things seem stable.

antillies wrote on 2021-09-24, 18:22:

Grem Five wrote on 2021-09-21, 20:51:

In my experience many weird errors can be traced down to power supplies. If you dont have clean power you will rarely have a stable system.

I know so many people that put tons of $$ into a system then put in the cheapest power supply they can find and then wonder why their expensive systems is so flaky.

I bought a Corsair CX450M 450W 80 Plus Bronze for this machine. Not the best, but also not the worst.

Would you have a recommended way to test whether the power supply is causing issues, besides replacing it? I'm certainly not opposed to upgrading it if the system continues to act up. As of right now, though, things seem stable.

I unfortunately dont have a good way to test for that, hard to tell if a power supply is suffering from bad ripple or transient spikes without some thing like an oscilloscope. Personally I have extra power supplies around that I can use to to see if the system runs any more stable.

Even very good built power supplies can vary from unit to unit, I got 2 identical Corsair PSUs that Johnyguru rated at 9.5+ and one is rock solid and the other has a 5 volt rail that is now all over the place and quite unstable, it still runs the pc but things can get very weird.

antillies wrote on 2021-09-24, 18:22:

I bought a Corsair... for this machine.

Corsair? This makes me feel like this cat: https://giphy.com/gifs/13Yok2fuVh4fcc

antillies wrote on 2021-09-24, 18:22:

Would you have a recommended way to test whether the power supply is causing issues, besides replacing it? I'm certainly not opposed to upgrading it if the system continues to act up. As of right now, though, things seem stable.

I examine power supplies before deciding about using them (or not) on my DIY test bench. Here a report with some discussion.
Reason is that bad PSUs cause so many problems and frustration that can be easily avoided.

I posted this link in another thread http://www.realhardtechx.com/index_archivos/P … ewDatabase.html its a good place to see who really made a power supply you are looking at.

If you look a company like Corsair they have had 6 different companies that have made their power supplies:

Channel Well Technology
HEC
Seasonic
Great Wall
Chicony Power Technology
Flextronics

I personally prefer seasonic psu, Johnnyguru was an awesome site to use to get an idea of how different psu performed but the site went down not long after he started working for corsair and I cant always find all the pages of some reviews on wayback machine. =(