VOGONS


First post, by subhuman@xgtx

User metadata
Rank Oldbie
Rank
Oldbie

Hello guys. I've been trying to set up a clean 98SE install in one of my built rigs for the past two days and -not-rarely- have come a not-so-litte inconvenience which is honestly driving me nuts. 😁

More specifically, a Kernel32 error which appears every single time Windows loads.

'Kernel32 caused an invalid page fault in module <unknown> at 0000:00000003'

The system in question is a 1400-S on an ASUS P3B-F 440bx board with 256 Megs of PC133 C2 RAM, and a 40gb IDE disk paired up with a Geforce3, a Matrox M3D in slot 2, a 12mb V2 in the third slot, an Intel 1000/GT nic in the fourth, a Montego Vortex 2 in the fifth and an AWE32 CT3980 which is in the bottom ISA.

Already tried removing cards one by one, re-installing Win with the /p i switch, running scanreg, reallocating IRQs and disabling programs on start-up with no success. The fun thing about this error is that most of the time I am able to continue straight to the desktop without any issue, but Oh yeah. Ever need to boot into safe mode and better forget about it. Error message, and bam that's it. Windows won't even attempt to draw the taskbar. 🤣

Enough rant, probably've been spending like 8 hours straight with this. 😵

Any suggestions? Thanks!

7fbns0.png

tbh9k2-6.png

Reply 1 of 8, by Jorpho

User metadata
Rank l33t++
Rank
l33t++
subhuman@xgtx wrote:

Already tried removing cards one by one, re-installing Win with the /p i switch, running scanreg, reallocating IRQs and disabling programs on start-up with no success.

Welp, that's all the obvious stuff.

How about memtest?

Reply 2 of 8, by firage

User metadata
Rank Oldbie
Rank
Oldbie

Kernel errors don't give a lot to go on. It could be hardware issues with memory, CPU or motherboard, it could be a faulty PSU causing instability in them, or just a driver error. Your 133 MHz bus configuration is higher than board defaults, and it runs AGP also above spec at 89 MHz, either of which could be a problem, although unlikely.

Did you try a reinstall with just the main display adapter and none of the other expansion cards or their drivers installed? That's quite a lot of stuff.

The interrupt sharing configuration for the board looks like: PCI 1 shared with AGP, PCI 2 not shared, PCI 3 shared with PCI 6 if present, PCI 4 shared with PCI 5 and USB. Most cards don't like to share; for the m3D an unshared slot is an explicit requirement, so its placement looks good. Voodoo2's can go anywhere - I'd put it in a shared slot rather than a sound card. Also, Vortex2's legacy emulation can cause trouble when running low on interrupts, I've had to disable that before.

My big-red-switch 486

Reply 3 of 8, by MrMateczko

User metadata
Rank Member
Rank
Member

Does this error occur when you have the first run 98SE after the install (Welcome to Windows 98 guitar solo!), or does it start to appear after you install a specific driver/change system settings/other stuff?

Reply 4 of 8, by subhuman@xgtx

User metadata
Rank Oldbie
Rank
Oldbie
Jorpho wrote:
subhuman@xgtx wrote:

Already tried removing cards one by one, re-installing Win with the /p i switch, running scanreg, reallocating IRQs and disabling programs on start-up with no success.

Welp, that's all the obvious stuff.

How about memtest?

Good idea, I had completely forgot about it. Left the rig running the tests for 11 hours: 21 passes with no errors.

firage wrote:

Kernel errors don't give a lot to go on. It could be hardware issues with memory, CPU or motherboard, it could be a faulty PSU causing instability in them, or just a driver error. Your 133 MHz bus configuration is higher than board defaults, and it runs AGP also above spec at 89 MHz, either of which could be a problem, although unlikely.

Did you try a reinstall with just the main display adapter and none of the other expansion cards or their drivers installed? That's quite a lot of stuff.

The interrupt sharing configuration for the board looks like: PCI 1 shared with AGP, PCI 2 not shared, PCI 3 shared with PCI 6 if present, PCI 4 shared with PCI 5 and USB. Most cards don't like to share; for the m3D an unshared slot is an explicit requirement, so its placement looks good. Voodoo2's can go anywhere - I'd put it in a shared slot rather than a sound card. Also, Vortex2's legacy emulation can cause trouble when running low on interrupts, I've had to disable that before.

Indeed, I am able to play pretty demanding games from 2000 at 1280x1024 without having the system crash once for hours (no matter if 100/133/150 fsb is used). Great to know about the resource layout. 😀 PSU in question is a 300Watter Delta DPS300MB which does 15A @+12v and 30A @+5v. Feels weighty and all. Should that suffice? Now tried to boot with nothing more than the Geforce3 without any seemingly difference. It is indeed quite a hefty setup and most of us here know by heart how fragile 95/98/ME are with hardware, so one cannot expect too much. 🤣

MrMateczko wrote:

Does this error occur when you have the first run 98SE after the install (Welcome to Windows 98 guitar solo!), or does it start to appear after you install a specific driver/change system settings/other stuff?

Apparently the error starts coming up after I run a program like Damn NFO viewer or Speedguide's TCP optimizer, though sometimes it does appear during the hardware detection process when running setup.exe over a previous installation. If it shows up and I hardreset the system without closing the window it won't appear till I don't run any 'conflicting' programs?, though if I don't It won't go away unless I re-format and reinstall good-ol Windows for the 10th time. 😵

The system is on a clean install once again. The same error tempted to show again with TCP optimizer but I could dodge the bullet by courtesy of a reset like said above. Any thoughts on this?

7fbns0.png

tbh9k2-6.png

Reply 5 of 8, by Joey_sw

User metadata
Rank Oldbie
Rank
Oldbie

try to active the bootlog option in win98SE boot menu.
after its crashed reboot to command line and see the bootlog.txt to find clue where its crashed.
looks for negative keyword such as: error, fail

-fffuuu

Reply 7 of 8, by leileilol

User metadata
Rank l33t++
Rank
l33t++

Those aren't mulitple video cards. It's one video card and two 3D cards 😀 I had a similar working setup sans Montego and it was on a 440LX w/ P2 at one point and didn't run into Kenrel32 issues.

I will at least mention that Nvidia Geforces do not get along with the PowerVR PCX1/2 drivers and will inflict all kinds of chaos and the solution to that is not use a Geforce (Rage, Radeon, 3dfx, Kyro, Riva128 and Matrox are known to be ok)

apsosig.png
long live PCem

Reply 8 of 8, by subhuman@xgtx

User metadata
Rank Oldbie
Rank
Oldbie

Weird, just finished installing all the drivers on the system, been using Opera 10.62 and playing games today, running the system at 100/133/150 fsb, no crashes at all. Deleted those two programs, as well as Rain 1.0, all good for now. Maybe they were infected indeed? Kind of voodoo magic' though. 🤣 Will report if I run into the same issue again, and how did it happen. To everybody, thanks for the advice & help. 😀

leileilol wrote:

Those aren't mulitple video cards. It's one video card and two 3D cards 😀 I had a similar working setup sans Montego and it was on a 440LX w/ P2 at one point and didn't run into Kenrel32 issues.

I will at least mention that Nvidia Geforces do not get along with the PowerVR PCX1/2 drivers and will inflict all kinds of chaos and the solution to that is not use a Geforce (Rage, Radeon, 3dfx, Kyro, Riva128 and Matrox are known to be ok)

Yeah I gotta give you the reason with that. With one my previous rigs, which had a Barton 2133 paired up with a KT333 and a Geforce2 GTS the system would often crash when doing crap like running 3DMark99MAX. Though weirdly enough on this Tualatin system I haven't run into any issues yet with the M3D. Maybe the incompatibility between Geforce cards and PCXs is caused mostly by driver revisions/chipsets and tends to affect apps running through the PowerVR D3D Hal the most? Mechwarrior 2 PVR works a treat, Tower4 and 5 tests do too.

For the record, Dx7a is installed and Detonator 22.50 is the driver being used.

7fbns0.png

tbh9k2-6.png