Not mentioned, but based on the Direct X versions you're using Windows 98? If it were me, I would put in a spare hard drive and try Windows 2000 or XP. This would either narrow it down to something wrong with the OS/drivers, or if there are issues with either of those operating systems then you likely have a hardware issue of some type.

Also, you ought to run DXDiag and see how far you can get with the tests.

To clarify, these are the boards you mentioned recapping? Have you ran memtest on them?

Repo Man11 wrote on 2024-03-19, 04:36:

Not mentioned, but based on the Direct X versions you're using Windows 98? If it were me, I would put in a spare hard drive and try Windows 2000 or XP. This would either narrow it down to something wrong with the OS/drivers, or if there are issues with either of those operating systems then you likely have a hardware issue of some type.

Yes Win98 SE.

I'll try Windows XP. Can I use an SSD with this? It currently has a Startech IDE to SATA for the 98 drive.

Also trying an FX5200 achieved the same result too. I just now tried. But didn't change drivers. Leaning on it being a driver issue right now.

I found the dxdiag program and it loaded up fine. No conflicts. And the direct3d test went without a hitch.

kingcake wrote on 2024-03-19, 04:55:

To clarify, these are the boards you mentioned recapping? Have you ran memtest on them?

They are yes. I have run memtest on them and they passed 5+ times.

less aggressive agp bios settings? or just failsafe settings across the board? i think you're using the same via 4-in-1 version i am if memory serves so not sure what else to try.

Did you replace all the big capacitors on the board (~470uF and bigger), particularly any near the AGP slot? For example, there's a picture of a K7T Pro2 here: https://theretroweb.com/motherboards/s/msi-ms-6330-ver-1.0a where you can see someone has replaced 3 big caps up by the CPU, but not the two by the AGP slot, both of which look like they've failed. Other than that, sounds like driver problems, I vaguely remember there was a correct sequence of the VIA, DirectX and Graphics driver installations.

Has your Thunderbird got all it's corners? Ppl used to report all kinds of weird crap happening back in the day for minor die chips... though "no noticeable problems" and completely inoperable seemed to also happen for what looked like the same amount of damage.

BitWrangler wrote on 2024-03-19, 15:48:

Has your Thunderbird got all it's corners? Ppl used to report all kinds of weird crap happening back in the day for minor die chips... though "no noticeable problems" and completely inoperable seemed to also happen for what looked like the same amount of damage.

Years ago I set some friends up with a Socket A system: Shuttle AK32A and a 1.4 Thunderbird. At some point it stopped working with anything 3D, and I suspected a problem with the AGP card. So I replaced it, and the problem remained. Then I suspected a problem with the motherboard's AGP slot, so I tried a PCI video card, and the problem remained. Then I replaced the CPU, and everything was fine. The Thunderbird CPU didn't have any visible damage, but for whatever reason it developed this oddly specific fault.

stanwebber wrote on 2024-03-19, 12:41:

less aggressive agp bios settings? or just failsafe settings across the board? i think you're using the same via 4-in-1 version i am if memory serves so not sure what else to try.

I'll give it a try. Doesn't seem to make sense being stable in OpenGL (Half-Life) but refuses D3D. The machines haven't thrown a BSOD or anything. Just D3D environment errors.

snufkin wrote on 2024-03-19, 14:52:

Did you replace all the big capacitors on the board (~470uF and bigger), particularly any near the AGP slot? For example, there's a picture of a K7T Pro2 here: https://theretroweb.com/motherboards/s/msi-ms-6330-ver-1.0a where you can see someone has replaced 3 big caps up by the CPU, but not the two by the AGP slot, both of which look like they've failed. Other than that, sounds like driver problems, I vaguely remember there was a correct sequence of the VIA, DirectX and Graphics driver installations.

I replaced the big 4. The 2 south of the CPU and the 2 east, not the ones at the top like the link. Those at the top for this board are the 'KZE' caps which appear to be fine. I have a few boards here with these KZE caps. None of them have visual issues and the boards are working well.
I also have a K7T Pro2 Turbo here. It has a different layout again of capacitors and is a red PCB. The little caps around the slots are the culprits on this one.

Might try a reinstall of 98 and sequence these driver installs. I believe most of what went down started after installing DX7.

BitWrangler wrote on 2024-03-19, 15:48:

Has your Thunderbird got all it's corners? Ppl used to report all kinds of weird crap happening back in the day for minor die chips... though "no noticeable problems" and completely inoperable seemed to also happen for what looked like the same amount of damage.

Now that you mention it, there is a slight nick off the corner of one CPU (K7T), but not the other (MK33).

I just found another palomino CPU here with no corners off it. Might give it a try. And a BIOS update for the K7T. Saying that the MK33 has the latest BIOS..

Repo Man11 wrote on 2024-03-19, 16:45:

BitWrangler wrote on 2024-03-19, 15:48:

Has your Thunderbird got all it's corners? Ppl used to report all kinds of weird crap happening back in the day for minor die chips... though "no noticeable problems" and completely inoperable seemed to also happen for what looked like the same amount of damage.

Years ago I set some friends up with a Socket A system: Shuttle AK32A and a 1.4 Thunderbird. At some point it stopped working with anything 3D, and I suspected a problem with the AGP card. So I replaced it, and the problem remained. Then I suspected a problem with the motherboard's AGP slot, so I tried a PCI video card, and the problem remained. Then I replaced the CPU, and everything was fine. The Thunderbird CPU didn't have any visible damage, but for whatever reason it developed this oddly specific fault.

I will try this, just to remove as a potential issue. I did try a different GPU (FX5200) with the same result.

LGR just released a video where he "restored" (LGR's version of restoration is a new CR2032) a Gateway and it ran OpenGL but crashed with DirectX. He ended up replacing the GPU after losing his mind trying different driver versions. Different GPU fixed it.

kingcake wrote on 2024-03-22, 16:57:

LGR just released a video where he "restored" (LGR's version of restoration is a new CR2032) a Gateway and it ran OpenGL but crashed with DirectX. He ended up replacing the GPU after losing his mind trying different driver versions. Different GPU fixed it.

It is doing it in windows XP (SP3) now too.. Same exact Unreal and UT99 issue. With the 128MB proper FX5200. I am so done with it.

But..

3DMark 01SE worked! But it scored a ridiculous 3790 3dmarks.. It chugged hard. Nature pulled 12.8fps.

The Pentium 3 733 815 is doing it AS WELL.

All different hard drives

If I change power supply and this fixes.. But a P3 733 too much for a n HX650W 5v rail??

I can't. There's another guy with an A64 doing the same thing on another site. All I have left to try is a Radeon. Totally different. Can't hurt to have one last go.