It should be a quality MBO with SANYO caps.
You can try modern ATX PSU with adapter. I think latest bios is a must for K6-2/3.

I do have the latest BIOS already so no issue there. I do have stuff here to make an ATX PSU adapter and I can do that to test but I have a limited number of ATX PSU's that put out enough amps on the 5v and 3.3v rails and need the ones I have for other builds ultimately but I was thinking after I initially posted that would at least be a way to verify

For reference, this is the Enermax AT PSU I have right now. Maybe someone see's something

okay, i don't think it's power supply. I wasn't metering it under load. the 12v holds steady at 12.3~v.

I've got the new k6 clocked at 333mhz right now and it's stable if I disable the external cache. I thought maybe it was the cache that couldn't handle the speeds but 333mhz became my new stable limit on the 500mhz chip.

Not really sure what else to look at now...

stlouis1 wrote on 2023-10-16, 13:40:

Aopen AP5T
AMD K6-2 400mhz
128MB SDRAM

1. You figured out the undocumented voltage setting, but you are still trying to run K6-2 in nonsuper7. I too had 430VX and TX boards behave very unstable with it, rebooting, etc. The opinion that I heard about it is that the power delivery circuits on boards like there are not designed for CPU as demanding K6-2. Midrange CXT top at about 20-21W TDP, compared this to Pentium 233MMX at 17W, and that is like the top thing, mind you. Common P166 parts were 10-15W. The speaking for itself analogy would be putting a 140W Phenom in a 125W rated motherboard. Back to the topic, do you notice the FETs get particular hot with K6-2?
2. Another thing to consider is flaky SD-RAM controllers in 430. If you have EDO/72pin or can at least use a slower module like a single 32mb PC-66, I would recommend it. Personally I had instability with as much as 64MB SDRAM on 430VX even though the system passed memtest86+.

Also, which exact revision you have? (1, 2, 3)

I/O voltage also powering a cache chips. make sure I/O voltage does not go below 3.3v or you cache chip will be unstable

Bruno128 wrote on 2023-10-17, 09:41:

[...]

2. Another thing to consider is flaky SD-RAM controllers in 430. If you have EDO/72pin or can at least use a slower module like a single 32mb PC-66, I would recommend it. Personally I had instability with as much as 64MB SDRAM on 430VX even though the system passed memtest86+.

For the i430VX I'd fully agree, but this is a 430TX with the same memory controler as in the 440EX. It can handle 256MB DIMMs (as long as they use 16Mx8 chips) and the only reason it doesn't do 512MB total is that it's been nerfed by Intel to avoid competing with more expensive platforms.

The compatibility isn't related to speed (RAM speed ratings are maximums, just like speed limits you can always go slower without ill efects) but to chip organization, as the i430VX - the first SDRAM chipset anywhere - had some pretty severe limitations. Newer DIMMs tended to have higher density chips, which the VX couldn't handle, older ones usually lower density stuff, but there was no hard guarantee that a PC-66 DIMM would have the correct configuration. For the TX, anything up to 128Mb per chip with at least 8 lines was fine (same as the venerable i440BX, except the BX supported more DIMMs and overall memory).

I left it running 3dmark 99 all night. Just came back this morning and it was still running. The FETs next to the CPU are pretty cool to the touch, at least the heatsinks are so I don't get the sense I'm straining anything there. Right now it's at 333mhz still with external cache disabled. I'd need to do more thorough testing but I'm not convinced disabling the external cache really did anything of it that was just a fluke on that particular run.

The motherboard's silkscreen says version 3.4

I'm using a double sided stick of 128mb PC100 and the bus is set to 66mhz so it is running slower. I've had the same issue with 72pin ram in there as that's what I was using initially.

The freezing always happens when trying to copy data from CD ROM to local disk, or installing a game from CD so I wonder if the issue is less with CPU or anything else. Maybe the issue something related to IDE controller specifically?

IDE controllers don't like being overclocked, but at 66MHz FSB / 33MHz PCI that shouldn't be the case here.

Are you running the HDD and CDRom on the same IDE channel or two different ones? If they're on the same, try separating them (HDD as primary master, CDRom as secondary master)

The drives are on separate channels. the IDE/CF reader is on the primary and CD ROM on the secondary as they should be

stlouis1 wrote on 2023-10-17, 11:22:

The freezing always happens when trying to copy data from CD ROM to local disk,

In Windows and in MS-DOS with OAKCDROM/MSCDEX?
Do you have another CD-ROM drive to test?

It's all in Windows. Haven't really used native DOS much for anything more than updating the BIOS. I do have another CD Rom drive actually. The one in the machine is an old Creative 52x. I've got an unknown 24-32x drive I could try, as well as some other more modern CD/DVD combo drives. I'm pretty sure I tested a different optical as I have one on the bench but I may have only used it on one of the other P3 / early athlon setups I'm still testing now that I think of it

Bruno128 wrote on 2023-10-17, 09:41:

You figured out the undocumented voltage setting, but you are still trying to run K6-2 in nonsuper7. I too had 430VX and TX boards behave very unstable with it, rebooting, etc. The opinion that I heard about it is that the power delivery circuits on boards like there are not designed for CPU as demanding K6-2.

Had a similar situation with a K6-2+ CPU on a Gigabyte GA-586ATV motherboard (based on the Intel 430VX chipset) where it caused restart loops during POST even after using a voltage adapter. The solution in my case was to downgrade the motherboard BIOS.

stlouis1 wrote on 2023-10-17, 11:22:

Right now it's at 333mhz still with external cache disabled. I'd need to do more thorough testing but I'm not convinced disabling the external cache really did anything of it that was just a fluke on that particular run.

The motherboard cache runs at the speed of the FSB and isn't affected by the CPU frequency.