More pics

We pulled it apart again and tried adding pull up resistors. But something seems to have happened that locked up the bios or something. It actually gave us quite a scare and we thought we broke it.

Turns out holding the sliding switch and pushing the blue power button a few times in a row did a reset or something and it fired up after putting it back to two shorted bf jumpers.

166 mhz.

I kinda think that maybe had we known it was going to do that that maybe one of the pull up settings maybe worked, but we decided to take the win and run with it. Put it back together and called it a day at 166mhz.

If we try again we will do it proper using a multiplier mini (see my pcb projects) and bodge wires to the multiplier mini. This will allow for actual dip switches and pull ups.

Re: Socket 5/7/SS7 (Multiplier Mini) Tweaker (Beta)

I have also overclocked my Armada 1550T to 166MHz, but I can only say one thing - CPU clock increase does not improve 2D graphics performance.

In general, the entire armada 15XX series suffers from slow 2D graphics. These laptops are not that much faster and sometimes even slower than a 75MHz P1 with asynchronous L2 cache... DIABLO lags on these laptops, even on a 1592 Armada...

Disabling spread spectrum speeds up 2D graphics better than bumping CPU frequency by an extra 33MHz. I tried to raise FSB frequency above 66MHz by experimenting with the FSB option resistors, but to no avail. Most likely clock gen. does not support FSB frequency higher than 66MHz.

And it's a pity, if it were possible to raise FSB to 75MHz, by a simple resistor reconfiguration, then, I believe, the performance of 2D graphics would improve drastically.

After OC, Vcore had to be bumped up by 0.02 Volts, otherwise applications, that required FPU, would crash.

SSTV2 wrote on 2022-12-18, 17:53:

I tried to raise FSB frequency above 66MHz by experimenting with the FSB option resistors, but to no avail. Most likely clock gen. does not support FSB frequency higher than 66MHz.

Where you able to find the datasheet for the MK1491-01 to ID the FSB resistors? Out of curiousity I had a quick look and I've only been able to find the -06, -09 and -14, and they're all very different (e.g. the -09 swaps the pin 1&4 Gnd and Vdd compared with the -06).

Can someone remind me if this cpu responds to pull up for 3x? I suspect my tests may have been false negative.

I have a 120 on socket 7 interposer that I tried to play with and see but couldn’t get it posting.

Not sure. I think the table ought to be:

1BF1 BF0 (pins 185 and 186)
2 0   1   x3
3 0   0   x2.5
4 1   0   x2
5 1   1   x1.5 (default)

but there's a note here: http://www.cpu-zone.com/Pentium/Pentium%20pro … ecification.pdf about a correction to the MMX datasheet regarding the x3 multiplier that

This bus ratio is currently not supported in mobile Pentium ® processor products.

Ah, just found what I think is the right datasheet: https://www.ardent-tool.com/CPU/docs/Intel/Pe … /242557-005.pdf and the BF table there just doesn't list 01 as an option. So at least according to the datasheet then it won't do x3.

So according to that we are indeed at the maximum frequency unless we can overclock the front side

I was trying to remember at what frequency the first generation socket 5 pentium CPUs ended

I guess it’s 2.5 X because I remember that they had a refresh of some sort that allowed them to get to 200 but I had thought that it ended at 133 and the new generation was 166 and 200 therefore I was thinking that this was of the new generation because it supported 166 but it appears to not be. Maybe there is overlap in any case it appears that the CPU only supports up to 166 MHz 2.5 X unless overclock in the front side

snufkin wrote on 2022-12-18, 23:02:

Where you able to find the datasheet for the MK1491-01 to ID the FSB resistors? Out of curiousity I had a quick look and I've only been able to find the -06, -09 and -14, and they're all very different (e.g. the -09 swaps the pin 1&4 Gnd and Vdd compared with the -06).

Unfortunately, I could not find the MK1491-01 datasheet. I found only one official document from the "MicroClock" manufacturer that mentions this generator.

In terms of leg placement, it mostly resembles MK1492-04 and MK1491-14, but is not a 100% match with either. I had left myself a note saying that pin 18 has a dual purpose of FS0 and PCI clock, thus I tried to look for the FS1 function among the other PCI clock pins, but unfortunately I had no luck finding it. I'm afraid that w/o lifting some legs it won't be possible to find the supposedly existing FS1.

Sphere478 wrote on 2022-12-19, 14:24:

I was trying to remember at what frequency the first generation socket 5 pentium CPUs ended

First gen. socket 5 pentiums ran up to 120MHz (1.5-2x).

This Compaq uses a 2nd gen. P54C, which theoretically should support 3x multiplier, though, I'd strongly advise you not to OC it that far, because of two reasons - heavily bottlenecked 2D and thermal CPU/cooling issues.

Re: Multiplier Bodge Board For laptops 486, Pentium, K6 , Cyrix.

Made this for it 😀

Sphere478 wrote on 2022-12-31, 22:44:

Re: Multiplier Bodge Board For laptops 486, Pentium, K6 , Cyrix.

Made this for it 😀

But the P54C's BF pins are internally pulled-up.

SSTV2 wrote on 2023-01-01, 17:08:

Sphere478 wrote on 2022-12-31, 22:44:

Re: Multiplier Bodge Board For laptops 486, Pentium, K6 , Cyrix.

Made this for it 😀

But the P54C's BF pins are internally pulled-up.

It’s universal.

I’ll take your word on the pull up thing for this cpu, but I know there are some pentiums 1s lacking it necessitating one to be added to some motherboards. Tillamook is one of these. Mmx might be the other?

P54 also doesn’t have bf2 🤣

If you look at the title of the thread it is for a range of processors

Btw, our measurements of the pads on that laptop showed those pins floating. It’s possible they default logically to high state but they don’t have a visible pull up from the pins. (If I recall correctly)

And double pulling high won’t hurt anything. At worst a unintended setting.

As long as you use high value resistors you are fine. If you don’t, you burn up resistors (or power supplies) but cpu would probably be fine. That pin(s) are supposed to read gnd or vcc3