I have a P200mmx that runs perfectly at 2.5x100 250mhz at default voltage.
There is a whole thread here somewhere all about this topic. I think it got as high as 300mhz but with higher voltage
and that was only with a real good cpu, most didn't reach that high.

I have an Overdrive 200MMX that overclocks to 250 no problem. But it's multiplier locked, so I run it at 83x3 (I didn't think it wise to push it to 100x3). I recently compared it to a K6-2 at speeds from 200-550Mhz. At 250Mhz, the MMX is roughly equivalent to a K6-2 300. My system is exclusively MS-DOS, so the MMX is a a great match. If I were running Win98, I'd probably prefer the K6-2 at 550Mhz.

Back in the day I had a Pentium 233MMX and it easily overclocked to 266, in fact I think I even got it to 290MHz (3.5x83) without any issues. Don't remember doing anything with voltage, but I may have.

Didn't get hot either, was a good/easily overclock.

mwdmeyer wrote:

Back in the day I had a Pentium 233MMX and it easily overclocked to 266, in fact I think I even got it to 290MHz (3.5x83) without any issues. Don't remember doing anything with voltage, but I may have.

Didn't get hot either, was a good/easily overclock.

Very encouraging! I think I found the thread referred to by Imperious ([http://www.vogons.org/viewtopic.php?f=46&t=50723) in which they talk about overclocking MMX chips on 100Mhz bus speed, at speeds up to 300Mhz, supposedly stable (although some using 3.1V which scares me a bit with a stock heatsink/fan).

I'll see if I can find a CPU intensive game that I'm comfortable playing for a while to test stability and try it at 75Mhz and maybe 83Mhz as you suggest. It would be very cool if I could get it up to 290Mhz using the stock heatsink/fan, but I suspect 2.9V might be required for that...

Like always with OCing. You should get a good sample that runs fast on low voltage. One of my P-233 goes up to 300MHz on default voltage and up to 315MHz on 3.1V. Some use even higher voltages.

RJDog wrote:

mwdmeyer wrote:

Back in the day I had a Pentium 233MMX and it easily overclocked to 266, in fact I think I even got it to 290MHz (3.5x83) without any issues. Don't remember doing anything with voltage, but I may have.

Didn't get hot either, was a good/easily overclock.

Very encouraging! I think I found the thread referred to by Imperious ([http://www.vogons.org/viewtopic.php?f=46&t=50723) in which they talk about overclocking MMX chips on 100Mhz bus speed, at speeds up to 300Mhz, supposedly stable (although some using 3.1V which scares me a bit with a stock heatsink/fan).

I'll see if I can find a CPU intensive game that I'm comfortable playing for a while to test stability and try it at 75Mhz and maybe 83Mhz as you suggest. It would be very cool if I could get it up to 290Mhz using the stock heatsink/fan, but I suspect 2.9V might be required for that...

What's the date code of your chip?

If I'm reading the production code correctly, it was made in 1997, 31st week. I guess not too long after the 233 was introduced...

Most should run fine at 262 MHz w/75 MHz FSB. I think running the PCI bus at 37.5 MHz is mostly OK. My chips also ran fine at 300 MHz, 3x100.

feipoa wrote:

Most should run fine at 262 MHz w/75 MHz FSB. I think running the PCI bus at 37.5 MHz is mostly OK. My chips also ran fine at 300 MHz, 3x100.

What's the date code of your chips? 😁

Tetrium wrote:

What's the date code of your chips? 😁

I assume this question is related to the idea that later chips (i.e. a time where Intel would be getting better yeilds) are more amenable to overclocking..?

RJDog wrote:

Tetrium wrote:

What's the date code of your chips? 😁

I assume this question is related to the idea that later chips (i.e. a time where Intel would be getting better yeilds) are more amenable to overclocking..?

Correct. It usually is, though there's often more to it then just the date codes.

The markings are:
FV80503233
L8180614-0338

How do you determine the datecode on these chips?

feipoa wrote:

The markings are: FV80503233 L8180614-0338 […]
Show full quote

The markings are:
FV80503233
L8180614-0338

How do you determine the datecode on these chips?

1998 week 18

Tetrium wrote:

Correct. It usually is, though there's often more to it then just the date codes.

Interestingly, I looked through my collection, and I seem to also have:

Pentium MMX 166 - 1998, week 1
Pentium MMX 200 - 2000, week 32

I'm inclined to think that the P-MMX 200 may be inclined towards overclocking, given its much later production date compared to the 166 and 233 that I have...?

RJDog wrote:

Interestingly, I looked through my collection, and I seem to also have: […]
Show full quote

Tetrium wrote:

Correct. It usually is, though there's often more to it then just the date codes.

Interestingly, I looked through my collection, and I seem to also have:

Pentium MMX 166 - 1998, week 1
Pentium MMX 200 - 2000, week 32

I'm inclined to think that the P-MMX 200 may be inclined towards overclocking, given its much later production date compared to the 166 and 233 that I have...?

That's the basic idea. However, when I tested all of my MMX chips I found that some MMX chips had some of their CPU multipliers disabled (particularly the later ones) which for obvious reasons give fewer overclocking options. None of the MMX 233 chips I tested had any of their multipliers disabled though.
You could simply first have your chip POST at the usual 200MHz and then try to have it POST at 233MHz, just to see if this multiplier is disabled or not.

I think a production year of 2000 for a desktop MMX chip is actually pretty late, most seemed to have been made in around 1998 or so?

I have a 166 mmx box (with the cooler from intel) and can go up to 300Mhz 99% stable

From my experience, it doesn't really matter the original speed of the cpu. I have several 166&200 models in my collection, and only one 233, and this last one is by far the worst when tried to overclock it. I really like these cpus as they match the one from my first computer, and usually buy them when find in thrift shops or sunday markets, and try to push them when at home 🤣 .

I've also approached that there's a difference between ceramic and plastic chips, which tend to overclock better than ceramic ones, but they usually have 'locked' multipliers. When I say locked I mean that the maximum multiplier allowed is the stock from that chip. Lower ones are always allowed (also keep in mind that MMXs translate 1.5x into 3.5x, which is only allowed on 233 models):

166MMX: 2.0x, 2.5x
200MMX: 2.0x, 2.5x, 3.0x
233MMX: 2.0x, 2.5x, 3.0x, 3.5x (same behaviour as 3.5x)

You can also find that when used in SS7 motherboards, almost all samples can do 250MHz (2.5x100) at stock voltage or 0.1-0.2V above 2.8V, and this really improves its FPU to fastest K6-2/3 levels in many games that are FPU demanding. However, reaching 300MHz is kind of difficult and leads to inestabilities in many samples of this chip (many can't even boot at this speed). It often requires greater voltages over 3V and the chip tends to run very hot, when achieved.

Here you are a very interesting post about this topic, very recommendable:
P55c at Overclocked to 300mhz?

So I tried clocking the P-MMX 233 (1997) at 262MHz (75 x 3.5) but the system wouldn't POST... not sure what's going on there, but ironically it would POST at 292MHz (83 x 3.5) and boy is it speedy.. noticable difference with RAM access, but for as speedy as it is, it is equally unstable. Fine at the command line, and 3DBENCH gave a result of 165.5, but load something a little more intensive up, like Doom or Quake, and... boom. Seg faults, random behavior, lock ups, etc.. This is running at the spec'd 2.8V, so when I get more time I might try again at 2.9V. I also want to try the P-MMX 200 (2000) to see if it might be any better.

Incidentally, the P-MMX 200 came out of some Cisco OEM equipment, which might explain the unusually late production date (assuming Cisco had a sourcing agreement with Intel) but could equally mean it has a fixed multiplier.

RJDog wrote:

So I tried clocking the P-MMX 233 (1997) at 262MHz (75 x 3.5) but the system wouldn't POST... not sure what's going on there, but ironically it would POST at 292MHz (83 x 3.5) and boy is it speedy.. noticable difference with RAM access, but for as speedy as it is, it is equally unstable. Fine at the command line, and 3DBENCH gave a result of 165.5, but load something a little more intensive up, like Doom or Quake, and... boom. Seg faults, random behavior, lock ups, etc.. This is running at the spec'd 2.8V, so when I get more time I might try again at 2.9V. I also want to try the P-MMX 200 (2000) to see if it might be any better.

Incidentally, the P-MMX 200 came out of some Cisco OEM equipment, which might explain the unusually late production date (assuming Cisco had a sourcing agreement with Intel) but could equally mean it has a fixed multiplier.

What board are you using btw? Maybe it's to do with the PCI bus being clocked higher at 75MHz FSB than at 83MHz FSB?
It's hard to tell without knowing your hardware configuration.