I have never used a single voltage Pentium MMX, but I recommend an equally interesting alternative, that is an AMD K6-233 which has a VCore of 3.2V by default, increasing it to 3.3V does not create any problem, the only precaution is to cool well both the CPU and the VRM mosfets, the latter being linear, emit more heat as the frequency increases, at 233 MHz they could be very hot.

I ran a Pentium MMX 233 in my Intel Advanced/ML motherboard for over a year with no issues. From what I understand most Pentium MMM CPUs can handle up to 3.7v, but that is probably pushing your luck, but 3.3v should be Ok.

They sometimes get gimped where you need the higher voltage to run stable.

You can typically drop your v/io as low as 3 volts and usually everything stays stable as long as you don’t overclock the bus speed.

PC@LIVE wrote on 2024-05-29, 18:11:

I have never used a single voltage Pentium MMX, but I recommend an equally interesting alternative, that is an AMD K6-233 which has a VCore of 3.2V by default, increasing it to 3.3V does not create any problem, the only precaution is to cool well both the CPU and the VRM mosfets, the latter being linear, emit more heat as the frequency increases, at 233 MHz they could be very hot.

That is an interesting idea and very valid considering the affordability. I wonder why it isn't talked about more as an upgrade path for socket 5 (assuming the board has dip switches that can take advantage of it).

dukeofurl wrote on 2024-05-29, 21:35:

PC@LIVE wrote on 2024-05-29, 18:11:

I have never used a single voltage Pentium MMX, but I recommend an equally interesting alternative, that is an AMD K6-233 which has a VCore of 3.2V by default, increasing it to 3.3V does not create any problem, the only precaution is to cool well both the CPU and the VRM mosfets, the latter being linear, emit more heat as the frequency increases, at 233 MHz they could be very hot.

That is an interesting idea and very valid considering the affordability. I wonder why it isn't talked about more as an upgrade path for socket 5 (assuming the board has dip switches that can take advantage of it).

If you have linear regulators it draws too many amps, I tried and the regulator got hot enough to fry eggs even with extra cooling then the system would crash.

dukeofurl wrote on 2024-05-29, 17:32:

I have two PCs, one is a gateway 2000 with a socket 5 Intel Zappa board, the other is an IBM PC 350 with a socket 7 board that d […]
Show full quote

I have two PCs, one is a gateway 2000 with a socket 5 Intel Zappa board, the other is an IBM PC 350 with a socket 7 board that does not provide the lower voltage for mmx processors.

I have been collecting anecdotes from various websites and social media of people upgrading these machines by running mmx CPUs such as a pentium 233mmx on each of these boards just as a drop in replacement knowing that the CPU is being over volted with the higher 3.3 voltage. While I don't have a big sample size - one anecdote for the socket 5 board, and maybe three anecdotes for the IBM board, it seems that each of these people have successfully run the mmx CPU in an over volted status for months without a serious issue.

I've seen plenty of warnings about how this can be bad for the CPU and cause it to break down over time, but on the other hand, I've heard anecdotes via Facebook groups of people basically doing this 20+ years ago (running a p233mmx on 3.3/3.3v) for years without issue and how that CPU at least anecdotally seems to be pretty robust for withstanding the added voltage.

So all in all, just wanting to broaden my anecdotal reports... Has anyone here done this (ran mmx CPU without split voltage), either finding that it was no big deal and the cpu performed fine, or finding that you burned out the CPU sooner or later? I've got these two machines without split voltage and it would be nice to totally max one or both out for low bucks considering that regular mmx CPUs are a tiny fraction of the cost of mmx overdrives...

Use a 233ANR AMD K6 or 200 MMX OD

I've never been a fan of putting a MMX at 3.3 or 3.5v. but yes, I have heard of people doing it. I think it's stupid.

Using Pentium MMX on 3.3v is absolutely fine and can't hurt it. It has the same fabrication node as the classic Pentium 200, which works at 3.3v. It works both ways too: nothing stops you from using classic Pentiums on 2.8v, if motherboard jumpers support such configuration.

PC@LIVE wrote on 2024-05-29, 18:11:

I have never used a single voltage Pentium MMX, but I recommend an equally interesting alternative, that is an AMD K6-233 which has a VCore of 3.2V by default, increasing it to 3.3V does not create any problem, the only precaution is to cool well both the CPU and the VRM mosfets, the latter being linear, emit more heat as the frequency increases, at 233 MHz they could be very hot.

Pentium MMX consumes less power at 2.8V, when compared to the classic Pentium lineup. At 3.3v 233Mhz it's would only marginally hotter than the Pentium 200.

The Serpent Rider wrote on 2024-05-30, 09:31:

Using Pentium MMX on 3.3v is absolutely fine and can't hurt it. It has the same fabrication node as the classic Pentium 200, which works at 3.3v. It works both ways too: nothing stops you from using classic Pentiums on 2.8v, if motherboard jumpers support such configuration.

PC@LIVE wrote on 2024-05-29, 18:11:

I have never used a single voltage Pentium MMX, but I recommend an equally interesting alternative, that is an AMD K6-233 which has a VCore of 3.2V by default, increasing it to 3.3V does not create any problem, the only precaution is to cool well both the CPU and the VRM mosfets, the latter being linear, emit more heat as the frequency increases, at 233 MHz they could be very hot.

Pentium MMX consumes less power at 2.8V, when compared to the classic Pentium lineup. At 3.3v 233Mhz it's would only marginally hotter than the Pentium 200.

Classic 200 ties vcc2 to vcc3 core is literally jumped to vio by the cpu

I saw this today re the amd k6 on socket 5 mb

Even though it runs at nearly standard voltage, the 3.2v/3.3v split voltage AMD K6-233 processor will likely have heat problems with these motherboards. Even with an adapter to adjust for the voltage and multiplier settings, the chip simply pulls too much juice through the socket and gives off too much heat to operate reliably. Although there are some Socket 5 motherboards with Socket 7 specifications (17 watts power dissipation, minimum 5 amps supply), it still isn't enough for the K6-233.

https://www.pchardwarelinks.com/cpu_sock.htm

It’s interesting, cyrix 166 was it? Had the crown as most TDP of any socket 5/7 cpu yet I pulled one out of a orange pc card the other day 🤣.

Note the size of the heatsink

OrangePC 530 pinout, bios update, vram upgrade

I’ve tried all sorts of crazy combos and struggle to think of an instance where anyone reported blowing up a regulator, I can think of instances of bad ones where people have replaced them, but no reports where: hey I tried 233anr or cyrix 166 and my regulator blew.

I’m sure it can happen, and one of you will reply with a link to one but it hasn’t happened to me in any of my crazy experiments. And none of that exact example come to memory.

Yeah, possible. Sure, Extra cooling sure, good idea. I say try it 🤣.

There are threads here about regulator upgrades too btw, you can upgrade your cpu power circuitry

vrm module project thread roundup, share ideas, make new designs

In my sig there are a few projects which allow injecting 3.3v directly into the socket, and/or cpu possible. If you wanted to re power the cpu, or supplement it.

This board...
https://theretroweb.com/motherboards/s/bcm-sq595
has a weak regulator, I recommend maximum of P166 classic, and even then make sure there's airflow over regulator.

I have blown the reg on that one, trying to get a K6 working, and I think I was only at 200Mhz when it blew... thought it was coping, though a bit hot, got into windows, ran some stuff and *poot* magic smoke gone. This was after putting bigger sink on reg.

edit: yah I think the guys who wrote the microhouse blurb knew, if you look at that, it only gives 150Mhz as highest setting, those came out the same day as 166, so that exclusion is a bit telling.

edit2: it's a bipolar transistor by the way, as current amplifier off real reg. Possibly you could find a more humungous one.

PC@LIVE wrote on 2024-05-29, 18:11:

I have never used a single voltage Pentium MMX, but I recommend an equally interesting alternative, that is an AMD K6-233 which has a VCore of 3.2V by default, increasing it to 3.3V does not create any problem, the only precaution is to cool well both the CPU and the VRM mosfets, the latter being linear, emit more heat as the frequency increases, at 233 MHz they could be very hot.

They're linear regulators, not mosfets. A mosfet is a single transistor. Inside the linear regulator are several transistors. And they're not FET, they're bipolar.

kingcake wrote on 2024-06-01, 06:59:

PC@LIVE wrote on 2024-05-29, 18:11:

I have never used a single voltage Pentium MMX, but I recommend an equally interesting alternative, that is an AMD K6-233 which has a VCore of 3.2V by default, increasing it to 3.3V does not create any problem, the only precaution is to cool well both the CPU and the VRM mosfets, the latter being linear, emit more heat as the frequency increases, at 233 MHz they could be very hot.

They're linear regulators, not mosfets. A mosfet is a single transistor. Inside the linear regulator are several transistors. And they're not FET, they're bipolar.

Thank you very much friend, unfortunately I am quite ignorant about the electronic components, but despite this, I can check them and understand if they need to be replaced, some time ago I repaired a MB S7 with a faulty linear regulator, only the single voltage one worked, after the replacement instead , dual voltage also works, and you can select 3.2V 2.9V and 2.8V.

I've found more anecdotes about people running a pentium 233mmx over volted on socket 5 boards without significant issues.

https://youtu.be/0SLIXEWy8l8?si=1vnz5Kfk41ApBR7X

https://forums.tomshardware.com/threads/old-s … ossible.492337/

Actually have one coming in the mail now so I'll put it to a similar test and see if my socket 5 system craps out or can remain relatively stable running things out of spec.

You could use an interposer. There is a build project in my sig. or they pop up on ebay every now and then.

That would definitely be a logical thing to do. I'm just not finding them for sale much, and when I do, they're priced at prices that cost more than what one of my PCs is worth. Not trying to brag but I have a skills gap when it comes to making my own 😅

Yeah, with cost of materials and what people make just going to their jobs it’s hard to justify making these projects for sale as people won’t make anything. You spend weeks gathering the parts, hours putting it together and random dude on internet says: eh, I’ll give you 20$ for it.

Why won’t anyone make and sell these?!?

Lol.

Yeah, the tools and instructions are there, gather the materials and pay someone to solder it if you have to. Some phone repair shops are good at stuff like this. Or you can try to find a interposer on ebay. They come up every now and then. Or you can build one 😀

When I designed it, I tried to make it as easy and cheap as possible to build to try and narrow the gap of those who might try to build it.

It’s really just a few pieces that need soldered together. And then you can use a off the shelf buck converter for power. Wasn’t able to make it simpler than that.

Just an update for some future internet person who stumbles upon this because they are thinking of doing the same thing:

-machine: Gateway2000 P5-75 with Intel Zappa socket 5 mother board, Gateway variant with no L2 cache, single voltage motherboard designed for pentium 1 cpus up to p133
-replaced cpu with regular socket 7 pentium 233MMX cpu, no overdrive, no interposer, added Socket 7 heatsink and fan combo
-motherboard does not go beyond a multiplier of 2 or a FSB of 65, resulting in a max supported cpu of a pentium 133, but setting the dipswitches to 1.5x66 for a 100mhz cpu triggers a multiplier in the pentium 233MMX cpu
-the bios regards it as a 100mhz cpu, but software and benchmarking programs appropriately recognize it as the correct cpu and there is a substantial performance boost in demanding games/benchmark programs so presumably the added cache and higher clockspeed of the cpu are working correctly
-the VRM on the motherboard did not reflect any physical issues while running with the mmx cpu
-the mmx cpu and heatsink were not particularly warm to the touch after using the system for several hours, in fact, the heatsink was much cooler than one I used with a p133 cpu on this board after running similar software
-so far I've used the machine in this configuration for about 4 hours consecutively, most of which was running demanding games to see what the performance boost was like. This is not a huge amount of time and I don't have any particular agenda to convince people to do this, so I'll come back and post again in the next days/weeks/months if the system starts to become unstable from the overvolting and/or suffers heat issues or some other issue. I'm not planning on running the system overnight or for 10+ hours straight to see if it crashes. I generally use my pcs for 1-3 hours at a time and then shut them down, so even if there were to be an issue from running for 10 or 24 hours straight, I'd probably never encounter it for the hobby usage that I want to use my PCs for.

I have a few ppga pentium cpus that turned brown from heat. How hot is yours getting?

I feel better saying this again, while I appreciate the mad science of OP I hope this doesn’t become accepted practice. This isn’t a good thing to do in my opinion. But I’m just some Schmuck on the internet. The proper way is Overdrive or interposer.