it sounds like the board simply lacks the jumper to set a 2x multiplier. and yes, it should work fine at 100mhz.

i think there are mods (bridging pins on the socket) to change the multiplier, if it's something you really want to do. and while i wouldn't worry about putting in a 133 with that, you need to be careful with a 200 on these old boards, because TDP is higher than the officially supported CPUs and it's not clear that the VRM is designed for it.

pentium 133 will be OK for this board，actually，the best cpu for this board will be PODPMT66X200

Thanks so much! Just tried it and the system booted just fine with this CPU, no magic smoke, but as expected it only has the CPU running at 100MHz. My prior understanding was that the clock multiplier was something internal to the CPU, not something set on the motherboard, so this makes sense... if I want to have this CPU running at 133MHz on this board it looks like I'm going to have to find a mod that'll set the clock multiplier to 2x.

Does anyone have any information on such a mod, or where to find info on how to figure out how to do such a mod, or whether it's even possible on this board?

Interestingly, looking at the manual again, I noticed that all of the jumper configurations seem to be for a 1.5x multiplier...

...so unless I'm missing something, the jumper settings seem to just be setting the system clock speed, not modifying the multiplier at all, so I don't know whether it'll be possible to run a 133MHz CPU on this board even with modification, because I have no idea how that clock multiplier is achieved, hardware-wise.

PlunderBunny wrote on 2024-11-20, 13:20:

Thanks so much! Just tried it and the system booted just fine with this CPU, no magic smoke, but as expected it only has the CPU running at 100MHz. My prior understanding was that the clock multiplier was something internal to the CPU, not something set on the motherboard, so this makes sense... if I want to have this CPU running at 133MHz on this board it looks like I'm going to have to find a mod that'll set the clock multiplier to 2x.

Yep. Locked multipliers internal to the CPU were a later thing; these CPUs need to be told how to behave.

Does anyone have any information on such a mod, or where to find info on how to figure out how to do such a mod, or whether it's […]
Show full quote

Does anyone have any information on such a mod, or where to find info on how to figure out how to do such a mod, or whether it's even possible on this board?

Interestingly, looking at the manual again, I noticed that all of the jumper configurations seem to be for a 1.5x multiplier...

20241120_080343.jpg

...so unless I'm missing something, the jumper settings seem to just be setting the system clock speed, not modifying the multiplier at all, so I don't know whether it'll be possible to run a 133MHz CPU on this board even with modification, because I have no idea how that clock multiplier is achieved, hardware-wise.

It's pretty simple really: socket 5 CPUs get their multiplier information from the BF0 pin. If it's high, you get 1.5x multiplier, if it's low, you get 2.0x multiplier. If the board completely fails to connect to it, the CPU pulls it up internally, so it becomes high and you get 1.5x.

So, to get 2x multiplier you need to pull BF0 low. Easiest way to do that is to solder a little wire to the relevant socket pin on the back of the motherboard and connect the other end of the wire to the nearest Vss (ground) pin.

See here for diagram of the socket and link to list of pin vs function allocations:
https://www.pchardwarelinks.com/586pin.htm

You could consider going a little bit further and also mod the BF1 pin (once again by connecting to Vss) to allow 2.5x and 3.0x multipliers. So7 CPUs would run at up to 200MHz. But... before you do that, you need to check if the linear voltage regulator on the motherboard is capable of delivering enough current at 3.3V for those faster, more thirsty CPUs.

dionb wrote on 2024-11-20, 14:27:

Yep. Locked multipliers internal to the CPU were a later thing; these CPUs need to be told how to behave.

Ah, okay, so I wasn't just making that up my thought process was just too modern, alright then!

It's pretty simple really: socket 5 CPUs get their multiplier information from the BF0 pin. If it's high, you get 1.5x multiplie […]
Show full quote

It's pretty simple really: socket 5 CPUs get their multiplier information from the BF0 pin. If it's high, you get 1.5x multiplier, if it's low, you get 2.0x multiplier. If the board completely fails to connect to it, the CPU pulls it up internally, so it becomes high and you get 1.5x.

So, to get 2x multiplier you need to pull BF0 low. Easiest way to do that is to solder a little wire to the relevant socket pin on the back of the motherboard and connect the other end of the wire to the nearest Vss (ground) pin.

See here for diagram of the socket and link to list of pin vs function allocations:
https://www.pchardwarelinks.com/586pin.htm

You could consider going a little bit further and also mod the BF1 pin (once again by connecting to Vss) to allow 2.5x and 3.0x multipliers. So7 CPUs would run at up to 200MHz. But... before you do that, you need to check if the linear voltage regulator on the motherboard is capable of delivering enough current at 3.3V for those faster, more thirsty CPUs.

Awesome, great info, thank you so much! Just to double-check, is that "BF0 pin high=1.5x, BF0 pin low=2x" standard absolutely universal for Socket 5, or is it something that could be different based on the motherboard? Just wanna double-check that before I make any modification.

...also, going by the above linked resource , if I'm reading everything correctly the BF0 pin for P54C CPUs would be the Y33 pin, I think? Could anyone who knows what they're doing better than I do confirm that for me before I start trying to solder things? 😆

PlunderBunny wrote on 2024-11-20, 14:52:

[...]

Awesome, great info, thank you so much! Just to double-check, is that "BF0 pin high=1.5x, BF0 pin low=2x" standard absolutely universal for Socket 5, or is it something that could be different based on the motherboard? Just wanna double-check that before I make any modification.

Behaviour of CPU under BF0 (and BF1 or even 2) is determined by the CPU, not the motherboard. All the motherboard does is pull those pins high or low (or in this case, nothing at all, leading them to read high). A P54C Pentium will interpret BF0 in the way I described. So will a Cyrix 6x86. An AMD K5 will ignore the settings entirely. Not sure about IDT Winchip.

And yes, BF0 is Y33.

Oh god right, good point, dunno what I was thinking considering that this is on the CPU, 🤣... great, I'll try pulling the BF0 pin on the socket low and see if that works, thank you again! 😁

After doing a surely beautiful and professional soldering job on the BF0 pin on the rear of the socket, the processor is now running at 133MHz! Or, well, 132MHz to be perfectly accurate, but what's a million cycles or so between friends? xD

Thank you again everyone for all the help!

Congrats and well done 😀

I have used a 233Mhz Pentium MMX on socket5 without a problem.

I was curious what the undocumented jumper config - JP6 1-2, JP7 2-3 would do, but you already got the desired result.

AlessandroB wrote on 2024-11-22, 20:14:

I have used a 233Mhz Pentium MMX on socket5 without a problem.

Two issues there:
1) you're overvolting the Pentium MMX from 2.8V to 3.3V. Those CPUs could handle a huge amount of abuse, but it's still not great.
2) bigger risk: you're drawing a lot more current over the 5V-3.3V linear voltage regulator on the motherboard than they were designed for.

A P233MMX draws 6A, whereas the fastest So5 CPU (100MHz) only draws 3A. That's asking for trouble with the regulator unless it's very over-specced for its original purpose (and well-cooled too).

BinaryDemon wrote on 2024-11-22, 21:19:

I was curious what the undocumented jumper config - JP6 1-2, JP7 2-3 would do, but you already got the desired result.

I was sorta wondering the same thing, but I've been too much of a coward to try it out... I wouldn't imagine a manufacturer would make an undocumented but easily configurable jumper setting that would melt a motherboard or anything, but I'm nervous to try it all the same. This computer is my aunt and uncle's old one from yeeeears ago that was just gifted to me so I really don't want to hurt it, and I'm confident enough with my soldering skills to make the clock multiplier mod to try that, but not confident enough that Gigabyte wouldn't do something really weird and potentially harmful with that jumper configuration. But hell, I should probably just try it, there's no way they would actually leave in the possibility for making such a simple mistake that could actually hurt anything, right???

AlessandroB wrote on 2024-11-22, 20:14:

I have used a 233Mhz Pentium MMX on socket5 without a problem.

Nice! Someday in the future I might get a faster CPU to put in there, but for now I'm happy with it as-is. Although now I wanna figure out how to actually make the turbo work, and how to reconfigure the cool 8-segment display on the front of the case to display the 133MHz CPU speed!

dionb wrote on 2024-11-22, 23:20:

A P233MMX draws 6A, whereas the fastest So5 CPU (100MHz) only draws 3A. That's asking for trouble with the regulator unless it's very over-specced for its original purpose (and well-cooled too).

It wouldn't actually be too difficult to replace the CPU voltage regulator with one specced to handle that much current, would it? And maybe do something to ensure it's properly cooled, like you said, but isn't that the only change that would need to be made?

PlunderBunny wrote on 2024-11-23, 10:26:

dionb wrote on 2024-11-22, 23:20:

A P233MMX draws 6A, whereas the fastest So5 CPU (100MHz) only draws 3A. That's asking for trouble with the regulator unless it's very over-specced for its original purpose (and well-cooled too).

It wouldn't actually be too difficult to replace the CPU voltage regulator with one specced to handle that much current, would it? And maybe do something to ensure it's properly cooled, like you said, but isn't that the only change that would need to be made?

But maybe the motherboard's copper traces may not be up to the task, actually damaging the PCB. Probably, the designers made it with most possible copper areas, but most possible isn't always enough, if doubling the current draw ...

PlunderBunny wrote on 2024-11-23, 10:22:

BinaryDemon wrote on 2024-11-22, 21:19:

I was curious what the undocumented jumper config - JP6 1-2, JP7 2-3 would do, but you already got the desired result.

I was sorta wondering the same thing, but I've been too much of a coward to try it out... I wouldn't imagine a manufacturer would make an undocumented but easily configurable jumper setting that would melt a motherboard or anything, but I'm nervous to try it all the same. This computer is my aunt and uncle's old one from yeeeears ago that was just gifted to me so I really don't want to hurt it, and I'm confident enough with my soldering skills to make the clock multiplier mod to try that, but not confident enough that Gigabyte wouldn't do something really weird and potentially harmful with that jumper configuration. But hell, I should probably just try it, there's no way they would actually leave in the possibility for making such a simple mistake that could actually hurt anything, right???

Probably safe enough to do it, as the jumpers are most likely configuration pins on the clock generator chip. You may want to search out the chip (probably close to the jumpers) and find a data sheet, to see what the fourth combination does ..

PlunderBunny, mind that this s.5 board doesn't support any dual voltage s.7 CPU. There's no option for adding a VRM module as upgrade, so it may successfully fry any dual voltage pentium MMX CPU even if you swap factory 3.3 V regulator with one for more current.

As for the undocumented bus frequency jumper combinations - they can't harm anything. In worst case there will be no POSTing with unsupported frequency combination.

As I suggested in other posts, if you want to run an MMX CPU at VCORE 3.3V, I would suggest the use of an AMD K6-233, which comes out 3.2V from the factory, the overvolt would only be 0.1V, but the talk changes on the Amperes, honestly I don't know if it absorbs too many, and if it puts the MB regulators at risk ⚠️, but I think that if the BIOS recognizes it, and if the temperature of the regulators remain within the limits, there could be good chances that it works, but wanting to use that CPU, you could try first at 133 MHz, and see if the BIOS recognizes it correctly, sometimes they are seen as 486DX-XX, but in any case the PC works, and through diagnostic programs, you can see both the correct CPU and the (exact) frequency.
The continuous use of a PC with such a K6 CPU, however, may not be advisable, rather it would be better to use a socket with an on-board regulator, with which you can select both the VCORE and the multi, for almost all S.7 CPUs.