The driver 1.07 does not have a good reputation. Try 1.06 or modern alternatives.

Сheck video memory using the program Video Memory Stress Test 1.7
If the video memory is OK, then most likely the problem is in the 3dfx chip itself

Thanks, hadn't gotten around to that test yet, but ran it on 2 different machines.

P3 1GHz
P3 450MHz (slot 1)

I did this to make sure it wasn't a problem with the newer 1GHz board which I haven't used that much yet, whereas my P3 450 is quite stable.

On both machines, I get the following error in the video memory stress test.
TEST FAIL (Code: 887601AE)

So something like the following error:
Trying 16bpp RGB:565 mode...OK
TEST FAIL (Code: 887601AE)
Trying 16bpp RGB:555 mode...NOT SUPPORTED (Code: 88760091)
Trying 16bpp BGR:565 mode...OK
TEST FAIL (Code: 887601AE)
Trying 32bpp RGB:888 mode...OK
TEST FAIL (Code: 887601AE)
Trying 32bpp BGR:888 mode...OK
TEST FAIL (Code: 887601AE)

Unfortunately, the documentation with the program doesn't help me with what exactly that error means. But I'm assuming it means there is some sort of issue with the SDRAM on the card?

WRT the driver version, the issues persistent with different driver versions, with the oldest 1.03 series exhibiting additional issues which appear to be strictly driver version related.

Anyone have a suggestion for next steps? I don't see any obvious damage to the board, but I haven't checked all the pins on the memory to make sure they are all firmly attached, etc.

Looks like it's time to look for a donor.

Ok, I stand corrected. I ran the video memory stress test on several of my video cards, including some that have never given me any trouble. And they all got the same error.

I read the documents on the test program again and noticed that I had missed this:

- Monitor standby function must be deactivated.

So it turns out that error occurs when the monitor is turned off on the system after 15 minutes! Hopefully this helps anyone else searching for the same error number!
You can disable this in Win98 be going to Control Panel->Power Settings and change the time out on the monitor to "never".

In any event, I ran the test again on the Voodoo 3 card and ran 22 passes in just under 2 hours with no errors.

I also ran 3DMark99 (DirectX 6 based I believe) and it ran 3.5 hours with no issues.

So...it's very odd to me that 3DMark00 (DirectX 7) is the only one that is exhibiting issues. Quake 2 (OpenGL) also seemed to work well as noted above.

Maybe it's just a specific feature that only DirectX 7 is using that occasionally glitches?

Does anyone know of another good DirectX 7 based demo or test program that I can try or have any other suggestions?

You can try the AmigaMerlin 2.9 driver, it's the best

Postman5 wrote on 2025-01-12, 13:23:

You can try the AmigaMerlin 2.9 driver, it's the best

Yep, I've tried that one, no change unfortunately. The particular issue I see seems to not be a driver issue.
I tried 3DMark2001 and noticed some very similar issues and even got an error about "CreateTexture failed: D3DERR_OUTOFVIDEOMEMORY", and I think that's a real error, but not my particular issue. Running at a lower screen resolution for it helped (at least for the brief testing I did).

I switched to Ultima 9, since it has glide and DirectX 7 support. And I noticed weird glitching artifacts on it too, seems worse at higher resolutions. And sometimes even had issues where you could just see though all the walls.

So seems like either a bad main 3dfx chip, maybe a cold solder joint? Isn't there an overclocking program, maybe I can try to mess with the voltages/clock to see if it stabilizes at all?

V.Control - you can try to lower the frequency
You can change the supply voltage by selecting resistors in the stabilizer. But this is not a radical solution to the problem. It seems that the 3dfx chip itself has degraded.

Okay, I think I've got it fixed now. I replaced all 6 of the 10uF 16V caps on the board (see thread here asking about IDing the parts: Voodoo 3 Capacitor Identification).

First, I tacked on 6 temporary 10uF 16V and saw that not only did the texture issues seem to go away...but 3D mark also ran approximately 2 times faster at the same resolution and settings. (i.e. double the frame rate. I compared with video recordings I did of the tests prior to the "fix"). I was not expecting that at all.

After that I went ahead and ordered 6 10uF 16V to replace the existing. Note that you can likely get away with a lower voltage but I was more focused on getting one the same size. Here is a link to what I replaced the old capacitors with https://www.mouser.com/ProductDetail/667-EEE-HD1C100AR Panasonic EEE-HD1C100AR

Use caution replacing the old ones as to not lift any pads but it wasn't too bad.

Seems to be working happily now!

I know this is a bit off topic, but I have an extremely similar system to yours. Same Motherboard, same CPU but I have a Voodoo 3 2000.

What BIOS are you running on the BX133-RAID motherboard? A custom or official BIOS?

Sleaka_J wrote on 2025-03-09, 13:46:

I know this is a bit off topic, but I have an extremely similar system to yours. Same Motherboard, same CPU but I have a Voodoo 3 2000.

What BIOS are you running on the BX133-RAID motherboard? A custom or official BIOS?

It's the official one, I've never tried messing with it. Apparently running a slightly older BIOS, not the lastest
ID string:
12/21/2000-I400BX-W977-6A69KA1CC-VI

According to The Retro Web (https://theretroweb.com/motherboards/s/abit-a … bx133-raid#bios), XK and ZW BIOS are newer, but I haven't tried installing them.

Guld wrote on 2025-03-09, 14:36:

It's the official one, I've never tried messing with it. Apparently running a slightly older BIOS, not the lastest ID string: 12 […]
Show full quote

Sleaka_J wrote on 2025-03-09, 13:46:

I know this is a bit off topic, but I have an extremely similar system to yours. Same Motherboard, same CPU but I have a Voodoo 3 2000.

What BIOS are you running on the BX133-RAID motherboard? A custom or official BIOS?

It's the official one, I've never tried messing with it. Apparently running a slightly older BIOS, not the lastest
ID string:
12/21/2000-I400BX-W977-6A69KA1CC-VI

According to The Retro Web (https://theretroweb.com/motherboards/s/abit-a … bx133-raid#bios), XK and ZW BIOS are newer, but I haven't tried installing them.

Ok. I've been looking for the custom 72 BIOS (it's at the bottom of the list you posted, but the date of it is later than ZW) that supports 48-bit LBA on the Highpoint 370 chipset. It used to exist 20 years ago, but every single link I've found to it is dead.

The best I've been able to find is a custom 48-bit LBA BIOS for the Abit BE6-II motherboard, which is apparently a close enough fit that the BIOS file works, but I know the proper one is out there.

I was hoping you had it. Alas, it was not meant to be.

Sleaka_J wrote on 2025-03-11, 11:51:

Ok. I've been looking for the custom 72 BIOS (it's at the bottom of the list you posted, but the date of it is later than ZW) th […]
Show full quote

Guld wrote on 2025-03-09, 14:36:

It's the official one, I've never tried messing with it. Apparently running a slightly older BIOS, not the lastest ID string: 12 […]
Show full quote

Sleaka_J wrote on 2025-03-09, 13:46:

I know this is a bit off topic, but I have an extremely similar system to yours. Same Motherboard, same CPU but I have a Voodoo 3 2000.

What BIOS are you running on the BX133-RAID motherboard? A custom or official BIOS?

It's the official one, I've never tried messing with it. Apparently running a slightly older BIOS, not the lastest
ID string:
12/21/2000-I400BX-W977-6A69KA1CC-VI

According to The Retro Web (https://theretroweb.com/motherboards/s/abit-a … bx133-raid#bios), XK and ZW BIOS are newer, but I haven't tried installing them.

Ok. I've been looking for the custom 72 BIOS (it's at the bottom of the list you posted, but the date of it is later than ZW) that supports 48-bit LBA on the Highpoint 370 chipset. It used to exist 20 years ago, but every single link I've found to it is dead.

The best I've been able to find is a custom 48-bit LBA BIOS for the Abit BE6-II motherboard, which is apparently a close enough fit that the BIOS file works, but I know the proper one is out there.

I was hoping you had it. Alas, it was not meant to be.

Good luck finding it! Upload it if you find it. Sorry I couldn't help!