What voltage does the CPU running at? Most DX2 parts were 5V ones but if it really runs at 100MHz it means that it has the 3x multiplier and is a DX4 indeed (it has 2x multiplier too - usually jumper selectable - so it can work as a DX2 too). What do its markings say?

There were 3V DX2s and I can easily imagine that these were actual DX4 parts 'binned' to DX2s.

If it's the case you should run the CPU at 3.3V instead of 5V but it must be printed on the CPU itself too. It should have an "&E3V1X" marking or something similar if it's an Intel part, simply "3 VOLT" if it's AMD, or something like "3.3 V" in case of a Cyrix.

Oh, and if it runs stable and at 3.3-3.45V it's perfectly safe to use at 100MHz (internal). At 5V it'll run much hotter than intended but may survive many years with sufficient cooling. I wouldn't bet my life on it though. Better safe than sorry.

There must be a FET close to the CPU that generates 3.3V from the 5V rail. Google showed that not every version of the board has it. And you must set it via the proper jumper too.

is your cpu from intel, amd or cyrix?
some dx2 cpus(most likely 3.3v ones from amd) may work at 3x clock if you set the clkmul pin to 3x.
also note that wt and wb 486s use different pin for clkmul. wt cpus use B13 pin for clkmul, which happens to be the pin that toggles cache wb/wt mode for wb cpus.

I had one like that too. It was some AMD DX2-66. But only the jumper for WriteBack cache turned the multiplier to 3X. I also tested with other DX2 cpus, but 3X only worked with that one.

It is Intel A80486DX2-66. It has that "&E3V1X" marking written on it.
I also checked voltage settings and apparently it is running at 3.15V.
I already tested it for several hours and it seems perfectly stable and cool with heatsink and a fan.
When I was trying to change multiplier from 3x to 2x via jumpers it was working at 66mhz, but also the system bus was underclocked to 22mhz. Weird...
Well, if you are saying that it should be okay, then I'll leave it as it is.

wow, never seen 3.3v 486dx2 from intel, i thought that intel only made 5v 486dx2.

Picture, please.

CPU:

BIOS detects it as:

It apparently does CPU auto-detection and fails on that, leaving the CPU overclocked. Check the BIOS if you can disable CPU auto-detection (unless you want to keep it that way).

Yeah, I' m keeping it that way. There is no such option in BIOS, the only way to set it is via jumpers (already set to 486DX2).

It may change if you set the L2 cache mode from wt to wb (or vice-versa).

There were 3V DX2s and I can easily imagine that these were actual DX4 parts 'binned' to DX2s.

I believe this AMD DX2-66 could be such example too. 16kb cache, write-back.

There are no L2 cache config jumpers on the motherboard. Changing cache mode does not change the CPU detected.

The Serpent Rider wrote:

There were 3V DX2s and I can easily imagine that these were actual DX4 parts 'binned' to DX2s.

I believe this AMD DX2-66 could be such example too. 16kb cache, write-back.

Nice, i have one of those. 😁

So I've tested Am486DX2-66V16BGC (which I've mentioned before) on the Luckystar LS486E board and looks like it's really relabeled DX4. In fact that's Am5x86 133ADZ (same thermal package) with locked 4x multiplier - jumper settings for Am5x86 133mhz reset it to DX2 66mhz.

Can you post pics of the motherboard as well?

I can post photos tomorrow, but why though? That's just common Luckystar rev. C2 board.

The Serpent Rider wrote on 2018-03-14, 14:13:

So I've tested Am486DX2-66V16BGC (which I've mentioned before) on the Luckystar LS486E board and looks like it's really relabeled DX4. In fact that's Am5x86 133ADZ (same thermal package) with locked 4x multiplier - jumper settings for Am5x86 133mhz reset it to DX2 66mhz.

This is interesting... so this CPU can run in 2x and 3x mode, but they've disabled 4x mode? In other words, normal AM5x86 133ADZ sets 2x as 4x, but on AM486DX2-6616BGC they've modified it in chip, so it runs as 2x?
I'm looking way to set regular AM5x86 133ADZ to 2x , can it be done somehow? Or is this behaviour hard-coded into chip?

No, I think it's not possible. 5x86s were highjacking DX4 multiplier changer, which is working like on/off switch. So it had to be done inside.