Has the mainboard been re-capped ? ie have its old capacitors replaced ?

I don't have the means to do so yet, but Windows XP appears to be perfectly stable and plays the example games and such, so surely that'd suggest that that isn't the issue, I'd presume.

You could try dropping the amount of RAM down to 512 MB or even 256 MB just to see if the issue stems from the fact that Win9x is not "supposed to" work with 1 GB. That's a clear separator between XP and 98 at least.

I did try with less RAM, as mentioned. Reducing it to physically 512MB and removing LIMITMEM. I didn't try 256MB but I've not had problems using 512MB before. Nor had I had issues with using 1GB and LIMITMEM. I also tried replacing the RAM with 512MB of entirely different sticks, but still had the same errors.

8b 50 04 8b 3d 50 32 03 10 74 03 87 c9 90 b8 39

Since your first two crashes happened on this same bogus code snip (except bytes 5-8 are different because there is an address relocation there), my guess is that it may be a corrupt DLL file. I don't know what would make the module name come up as a temporary file. But if you search your HDD (use XSearch or something) for a file containing some of these bytes (eg. bytes 9-15) you might be able to identify the file.

Are you using any of the unofficial Service Packs for Win98SE? Also, which DirectX version are you using?
If I may also suggest removing the USB2.0 card entirely, at least for a while, to observe if this behavior stops occurring.

In the interest of ruling things out, I tried a different graphics card, my spare "test" Fx 5200, just in case maybe the RAM on the graphics card had an issue, but of course, no change.

bakemono wrote on 2022-12-30, 10:59:

8b 50 04 8b 3d 50 32 03 10 74 03 87 c9 90 b8 39

Since your first two crashes happened on this same bogus code snip (except bytes 5-8 are different because there is an address relocation there), my guess is that it may be a corrupt DLL file. I don't know what would make the module name come up as a temporary file. But if you search your HDD (use XSearch or something) for a file containing some of these bytes (eg. bytes 9-15) you might be able to identify the file.

I tried searching the hexadecimal with XSearch but it didn't come up with anything. I tried searching "exact phrase" for the first or last five (the ones common to the first two errors in my original post), but XSearch doesn't do anything, and so I searched for "all of the words" and it comes up with COMMAND.COM, IO.SYS and LOGO.SYS, but those just seem to have isolated instances of each hex pair (like, happening to contain one 8B somewhere).

Garrett W wrote on 2022-12-30, 11:17:

Are you using any of the unofficial Service Packs for Win98SE? Also, which DirectX version are you using?
If I may also suggest removing the USB2.0 card entirely, at least for a while, to observe if this behavior stops occurring.

The only unofficial Windows 98 extension thing I've been using is the unofficial USB driver. My DirectX is 9.0c.

I've removed the USB card, disabled the onboard USB ports, removed all the USB devices in Device Manager, and uninstalled the unofficial USB driver. I still get the error. However, it is slightly different now, though contains the same byte snippet.

1IC caused an invalid page fault in
2module ~DECC24.TMP at 0187:0209699e.
3Registers:
4EAX=020d1ff9 CS=0187 EIP=0209699e EFLGS=00010206
5EBX=00b9f234 SS=018f ESP=00b9f0ac EBP=00b9f0cc
6ECX=00000000 DS=018f ESI=00000000 FS=18ef
7EDX=81bd1971 ES=018f EDI=020d1ff9 GS=0000
8Bytes at CS:EIP:
98b 50 04 8b 3d 50 b2 0a 02 74 03 87 c9 90 b8 39 
10Stack dump:
11020d1ff9 022093a3 00b9f234 fe55cef8 244cf433 3fab1a32 9a44822d 9e3779b9 00b9f118 02096b27 020d1ff9 020b7460 020d09ac 020d0ef9 02099b9c 020d0ef9 

(I have since earlier restored an earlier installation, from when I know things had been working, since the freezing on changing the resolution was clearly not healthy. This earlier installation doesn't freeze when changing resolution, at least, though the other issue is still present (as you can see). Not sure if this caused this minor change in error or the removal of the USB things, though it's still the same error fundamentally, so it's probably not consequential.)

Which USB driver would that be? nusb3.3 ?

If I may suggest something, since I understand you're dual-booting WinXP along with 98SE and because DirectX versions beyond 7.0a have been known to cause weird issues under 98SE, would you consider keeping 98SE strictly at 7.0a?
I would also suggest removing that USB 2.0 card all together, they are more trouble than they are worth in my opinion. Internal USB usually works fine, especially on a system as fast as yours, but if you want more speed and immediacy, you already have a network card in there which you can use, I personally have an FTP server setup on my main rig and have retro systems connect to it and grab files. I am not a networking guy at all and found this an absolute breeze to set up a couple of years ago, to the point where I've disabled USB completely on all my retro systems running Win9x.

Also, there is actually one unofficial Service Pack that you should perhaps consider installing, and that's 2.1a. If anything, because it is likely the easiest way to install all official Win98SE updates in one go. It has a few (selectable) tweaks and utilities as well and it changes some GUI elements with ones from Win2000/ME (regrettably) without an option to avoid doing so. It's worth it IMO.

Finally, what networking card are you using out of curiosity?

Alright. I've had some time to do more troubleshooting.

- I've taken all the expansion cards out of this system except for the FX 5700. Also removed the 512MB RAM stick, leaving 512MB in there.
- Did a completely fresh install of Windows 98.
- I've installed only the drivers for the chipset (Apollo Pro 133T), graphics (Nvidia 53.04, then tried updating to 81.98 with the same result), and DirectX 9.0a. Nothing else. No USB drivers, sound card, network or anything (and the motherboard's USB is still disabled)
- I then installed the game I'm using to test this, Impossible Creatures, and...it worked! It all played just fine. I went into the campaign and loaded a map, played around a bit for a moment and all was well. I exited the game, feeling ready to go on with putting cards and things back into this to see which had the problem.
- I decided to not rest on my laurels, and double check. I immediately went to start up Impossible Creatures again; the last thing I did was quit the game, I'd not shut down or anything. However, this time...the error came up again. It looks a bit different, with 0167 this time, but it's still the same DECC24 file mentioned. A Kernel32 error came up immediately after exiting the error one time though.
- I installed and tried Rome: Total War as the other game that was having the error (and even updated DirectX to 9.0b that came with it), and as expected, the error occurred as it had done previously.

I then decided to put all the cards and RAM back in and change over to XP, and sure enough, these games still worked perfectly fine on there.
(And no, this isn't a dual-boot, this is completely isolated on a separate hard drive, and I just swap the cables over.) a sound or network card issue. I've also previously tried a different graphics card, RAM and hard drive, and still got the same error.

I didn't try installing a lower version of DirectX, but from my limited testing of games it really only seems to be in games which need DX8 or later that have the issue, and even then not all of them by any means. For example, Jedi Outcast still works just fine on this system in Windows 98.
I should maybe also mention that I did get a similar-looking error with the basic version of Anachronox, which uses OpenGL instead. However, patching the game fixed that, so it stopped being a practical issue, and I'm not sure if it even would be related. I'll try it again at some point perhaps, though.

And as well, part of what I had been using this system for was these end-of-era Windows 98 games requiring DX8 or 9, regardless of actual performance. I have other systems for older games, and wanted to be able get the most out of the Tualatin and Geforce FX 5700, even if they struggled a little bit.

TL;DR:
I'm still getting this error. After a lot of testing and fresh installs, I can safely say it's not a USB issue, nor is it a video driver issue, or a sound or network card issue. I've also previously tried a different graphics card, RAM and hard drive, and still got the same error.
This system had been running these games just fine, without this error, until recently. It doesn't seem compelling to me that this is some kind of inherent incompatibility or quirkiness in Windows 98.

Things that still seem plausible to me:
- A hardware issue. Perhaps there is something going bad about this motherboard, or maybe even the CPU. It's just really strange that the games still run completely fine under Windows XP, but perhaps there's some way that they interact with the hardware differently when running exactly the same games?
- Maybe a DLL did get corrupted, in the DirectX folder I always use to install it from, but I'm finding this quite unlikely in the scheme of things. Sometime I'll try and install it directly from a disc just to completely rule it out. As mentioned earlier here, I did update DirectX with 9.0b from Rome: Total War with no change.
- Maybe there's some kind of resource issue occurring that's hard to notice? I wouldn't expect that kind of thing on such a minimal install as I've just done, however. One single expansion card for video, I might have had one COM port enabled, but no USB or anything beyond that and other essentials. Not even sound, and definitely not ISA sound.

Feeling like I'm approaching a frustrating dead end. Fortunately I can still (for now?) play many of the games I'm wanting to on here, but it's a real big head-scratcher that these certain games that had been working now aren't.

I hope this additional testing and information might help narrow some things down. I appreciate all the help even if it seems like I've been dismissing a lot of it so far.

The system hang is a bit perplexing, but I have a few follow-up questions:

a) which Apollo Pro 133T drivers are you installing under Win98?
b) at which point during this testing did you install Nvidia drivers 81.98? Later drivers (I forget if 81.98 are the last ones) are notoriously broken under Win9x.
c) you say that this was a non-issue until recently, did you change your Nvidia drivers recently or were you always on 53.04?
d) lastly, have you tried doing a BIOS reset to defaults while keeping conservative settings?
setting your RAM to CL3 and perhaps even PC100 speed rating and underclocking your videocard in both core and memory a little bit would help make 100% sure the problem does not lie in one of these components
You never know, XP being NT based is far less likely to have an instability take the whole system down, compared to anything Win9x. I'd give it a try.

Also, I'm not sure you've answered my previous question. The scenario in which the Win98SE setup is aimed at titles up to and including DX7 games, and WinXP takes care of anything above and beyond that, is that unappealing to you? It could lead to less issues down the line, even if this particular issue you're experiencing is proven to be unrelated.
I should add that AFAIR Aquanox is a DX8 game and Aquanox 2 DX9. Jedi Outcast is an OpenGL game, it's running on idtech3, the engine that powers Quake 3.

Ollydbg works in Win98, maybe try launching the game with it and see if any useful hint comes up when the error occurs?

The way that the error is occuring in a temporary file definitely points towards a software issue, not hardware. Something seems to be injecting itself/hooking into other apps - maybe a virus, or maybe something by design e.g. a third party mouse util or other background app. Grab a copy of the referenced tmp file and examine it to figure out exactly what it is.

Alright everyone

I've found the cause of the problem.

I admit, I hadn't touched the settings in the BIOS after "optimising" them some time ago. Because, the system itself didn't seem erratic, "unstable", and nor did they seem too much (RAM CL was still at 3 for instance), and perhaps I'd also changed something there and didn't test thoroughly before forgetting. But I had a bit more time today to sit down and reset and comb through them.

Enabling optimisations one-by-one and testing each time, I came upon the setting causing the issue: "IDE Secondary Prefetch".

I figure this enables prefetching for the secondary IDE controller? Perhaps that doesn't work for the CD drive, or the CD drive in this PC, or something. Anyway, all my other optimisations seem to work now, and so do these games.

Garrett W wrote on 2023-01-04, 23:50:

Also, I'm not sure you've answered my previous question. The scenario in which the Win98SE setup is aimed at titles up to and including DX7 games, and WinXP takes care of anything above and beyond that, is that unappealing to you? It could lead to less issues down the line, even if this particular issue you're experiencing is proven to be unrelated.
I should add that AFAIR Aquanox is a DX8 game and Aquanox 2 DX9. Jedi Outcast is an OpenGL game, it's running on idtech3, the engine that powers Quake 3.

I wanted to use this hardware as a Windows 98 "powerhouse" to play the end-of-era stuff on, while still providing much of the typical Windows 98 features such as native DOS backwards compatibility (for the less-sensitive titles), with only perhaps changing the FSB being necessary. The Tualatin is a Pentium III, so being a powerful Windows 98 system feels much more fitting for it than being a really low-end/early Windows XP system. I just put XP on because I could and I was curious, and there were a handful of games that ought to work on 98 that have some other strange issues, that I still wanted to test with the hardware. I have other systems that are more suitable for XP.
Also, I was talking about Anachronox, not "Aquanox". And, Jedi Outcast lists DX8 as a requirement on its box, and that's what I was going off of. Perhaps that just provides a necessary OpenGL driver, too.

bakemono wrote on 2023-01-05, 12:21:

Ollydbg works in Win98, maybe try launching the game with it and see if any useful hint comes up when the error occurs?

I've not heard of that, and perhaps I'll give it a try sometime to see why this BIOS setting causes this problem, if no one else would have any clue.

jmarsh wrote on 2023-01-05, 17:39:

The way that the error is occuring in a temporary file definitely points towards a software issue, not hardware. Something seems to be injecting itself/hooking into other apps - maybe a virus, or maybe something by design e.g. a third party mouse util or other background app. Grab a copy of the referenced tmp file and examine it to figure out exactly what it is.

It was really strange, when I tried to search for the file name, nothing showed up at all. I have hidden files set to be shown and everything.

But, at least I seem to have found the cause, even if I have no idea why it's causing the issue. I'd be curious if anyone has any suggestions for what could be going on there.
Thanks for the input everyone.

These "temporary file names" remind me of some CD copy protection schemes, e.g. SafeDisc/SecuROM. I wouldn't have paid much attention to it, but since you mention the problem is related to an IDE controller setting...

It could indeed be that "IDE Secondary Prefetch" was breaking SafeDisc/SecuROM somehow... And that would explain why it only seemed to be the small niche of later Win98-compatible DX8/DX9 games, since, from what I understand, that's the era it started getting implemented.
I was checking DXDiag, and that was all working fine, and the error was occurring right on starting the game.

I don't know enough about CD DRM methods to know why prefetching would break them, but it seems quite probable now that you suggest it.