Aside from this specific motherboard case, does anyone here have experience or tips with running Pentium MMX chips at 3.3V?

Aside from running hotter, I'd think they'd tolerate the overvoltage reasonably well given the 350 nm process used and that the 350 nm Pentium Classic, Winchip, and K6-233 all ran at 3.3 and/or 3.52 volts.

They'll work on 3.3V, and the maximum spec voltage is actually 3.7V (though that's the stress rating rather than the functional rating). But you'll definitely want some decent cooling on it, and even so I don't know what the long-term reliability is like. Pentium MMX chips are cheap and plentiful, though, so it's not a huge loss if you fry it.

There are a couple other options too, if you want to play it safe... Intel made PMMX Overdrive versions, which were specifically intended for use at 3.3V. Also, a few companies made voltage converter/adapters that were basically voltage regulators that went in between the CPU and socket. Either one can be found relatively frequently on eBay, for fairly reasonable prices.

Old Thrashbarg wrote:

They'll work on 3.3V, and the maximum spec voltage is actually 3.7V (though that's the stress rating rather than the functional rating). But you'll definitely want some decent cooling on it, and even so I don't know what the long-term reliability is like. Pentium MMX chips are cheap and plentiful, though, so it's not a huge loss if you fry it.

There are a couple other options too, if you want to play it safe... Intel made PMMX Overdrive versions, which were specifically intended for use at 3.3V. Also, a few companies made voltage converter/adapters that were basically voltage regulators that went in between the CPU and socket. Either one can be found relatively frequently on eBay, for fairly reasonable prices.

I picked up a couple MMX 200s and a 233 for about $3 each, so I wouldn't really mind risking it.

The main concern (for bot me and the friend in question) is whether there's any risk to damaging the board, namely whether the voltage regulator can handle the added power consumption.
However, since several of the socket 5 native 6x86 chips should be similarly hot/power hungry (if not more so), then that shouldn't be an issue either. Albeit, it is an intel-specific OEM board and such parts many not be taken into account. Given the apparent fixed 3.3V setting, it would also bar the many 6x86/K5/Winchip parts requiring 3.52V.

Plus, unless those adapters and PODs are using nonlinear regulators onboard, the power consumption should be very similar to running a P55C natively at 3.3V. (a linear regulator would simply be cutting that 3.3V down to 2.8 and dissipating the heat -vs nonlinear converting voltage and amperage)
The board seems to be cleared for the POD MMX 200, so running a plain MMX 200 at 3.3V would probably be the safer limit rather than attempting a 233. Though that old benchmark archive mentioned in the opening post specifically mentions using a 233.

As for cooling, the current system is already set up with a decent heatshink and fan with thermal paste applied. The heatsink is on the small side, but not much different from what's common on PGA Mendocino Celerons and a fair bit heaftier than the stock Intel heatsinks for Pentium MMX.
That set-up is probably overkill for the Pentium 133 currently in the system. It seems it was pretty common to not even bother with thermal pate for those CPUs, or even no fan if the heatsink was a decent size and a case fan present.

Well, you'll really need to look at the board's voltage regulator chip to be sure on that. Just find the number on it and do a search for the datasheet... that should tell you the rated amperage on it. If it's good for 7A or more, you should be fine, 6A would probably do it but you'd want to keep a close eye on it to make sure it's not getting too hot. I wouldn't try it on a <6A regulator.

And you really can't judge based on the power consumption of 6x86 chips and the like. Remember, this is an old Intel board you have... it shouldn't be much of a surprise that Intel boards have rather poor support of third-party processors.

Old Thrashbarg wrote:

Well, you'll really need to look at the board's voltage regulator chip to be sure on that. Just find the number on it and do a search for the datasheet... that should tell you the rated amperage on it. If it's good for 7A or more, you should be fine, 6A would probably do it but you'd want to keep a close eye on it to make sure it's not getting too hot. I wouldn't try it on a <6A regulator.

In that line of thought, it would be good to make sure the regulator also has a decent heat spreader on it and not a barely-adequate tiny cheap one that some tend to use.

And you really can't judge based on the power consumption of 6x86 chips and the like. Remember, this is an old Intel board you have... it shouldn't be much of a surprise that Intel boards have rather poor support of third-party processors.

Right, I mused on that possibility above too, and the fixed 3.3V setting is pretty indicative of that too.

I'm the friend kool_kitty mentioned, and there's no problems to report so far. It seems to run pretty cool even though it's not running off split-rail voltage and I managed to get a nice little speed boost out of my DOS/Win 95 rig.

Oh, cool, you can log-in again. 😁

For those interested, here's the original thread for Apolloboy's Pentium build:
My Pentium 133 build

As you can see in his sig, he's since upgraded to 128 MB EDO DRAM, Win95, and a Velocity 128 (Riva), and of course the MMX 200.

He's also running it through an external video scaler/converter box to a PVM monitor through S-video (cleaner and less buggy than the Riva's onboard TV-out, or the ISA TV-out card he was using before -which was also DOS-specific iirc). So practically limited to ~640x480i max resolution. (and 60 Hz interlaced refresh) Still, not bad for what the sort of games the is being used for, and certainly not bad considering console gaming standards of the same period.