RayeR wrote on 2023-11-16, 05:27:

... I wonder what BIOSes and MBs use people whose got Tillamook detected as "Tillamook" - some custom tuned BIOS?

I'm running mine on a Chaintech CT-5RSA0 @ 95MHz x3.5.
Vio tuned to ~3.1V (lower will make system unstable) and the usual Tillamook mods for L2 cache/BF2/etc.
BIOS is Jan's K6plus-modded one as far as I can remember. Nothing special at least. But the stock one should have Tillamook support too since Jan only added K6plus optimizations afaik.

After some issues with further internal shorts and milling more holes and fixing all the mess I finally sealed it by epoxy and soldered the copper IHS back on the top side of interposer. I used my weller solder iron where I simply removed a solder tip and put the IHS to front side of solder pen, turned temp to the max and waited unit it melted and then power off. I also soldered the necessary bridges for voltage and cache (BF2 already set by 0R on bottom side).
Now doing some test at Gigabyte GA-5AA, I solved cache issues by flasing latest beta BIOS F7, see here:
Re: Tillamook 266MHz and working L2 cache?

Very nice! :p

This is no small accomplishment those are very very delicate chips!

Very nice ! This makes me wonder if I should extract a 300MHz kind from a laptop I have... 🤣

Tiido wrote on 2023-11-22, 23:26:

Very nice ! This makes me wonder if I should extract a 300MHz kind from a laptop I have... 🤣

Is that laptop functioning by chance? If so, would you mind runng a couple CPU information utilities on it? And if possible, probing bf0,1,2 pull up/down state with DVM?

It definitely works, although it is packed away and not outright accessible. It is a Toshiba Portege 3020CT

Tiido wrote on 2023-11-23, 05:16:

It definitely works, although it is packed away and not outright accessible. It is a Toshiba Portege 3020CT

I have been trying to verify the existence of the tillamook 4.5x multiplier for some time. I am beginning to wonder if it even exists? I am wondering if those laptops maybe had 4x 75mhz =300?

I have found this specs https://usermanual.com/support/toshiba/docume … t-specification that claims 300MHz at 66MHz FSB so there should be Tilly version with 4.5x multi. I did not find any newer datasheet from intel mentioning this version but intel probably delivers it only for corpo customers and not end users so the specs might be under NDA.
If the laptop is in good condition I wouldn't destroy it just for the CPU, mine was a scrap mechanically damaged...

Sphere478 wrote on 2023-11-23, 05:20:

Tiido wrote on 2023-11-23, 05:16:

It definitely works, although it is packed away and not outright accessible. It is a Toshiba Portege 3020CT

I have been trying to verify the existence of the tillamook 4.5x multiplier for some time. I am beginning to wonder if it even exists? I am wondering if those laptops maybe had 4x 75mhz =300?

CPU-z (vintage 1.04) doesn't really know what to make of it, any other utilities I could try?

Speedsys knows what it is though 😀

This 3020CT laptop is currently assembled and working, but could with some effort be unassembled for measuring pins under the microscope, if you can point to where I should check.

edit: I had no idea the 300MHz was pretty rare, looks like this laptop (3020ct & 3025ct), the Satellite 2530/2535 CDS/CDT and some Acer-made Thinkpads, the i-series 1400 (2611)'s CPU card (10L1268) seems to be where most of the pictures online came from.
And the HP5-MX300-L 😀

Well, that seems to confirm it,

Super curious to know the bf configuration

I would bet it's BF2:0 = 1 0 1, so let some owner try to measure it on his laptop. But I'm not sure how valuable this info would be as lower CPUs are already locked. Intel don't give bonus easily as AMD. AFAIK nobody was able to unlock multi on intel there was only some unlocked engineering samples, the same for SMP...

RayeR wrote on 2023-11-24, 00:00:

I would bet it's BF2:0 = 1 0 1, so let some owner try to measure it on his laptop. But I'm not sure how valuable this info would be as lower CPUs are already locked. Intel don't give bonus easily as AMD. AFAIK nobody was able to unlock multi on intel there was only some unlocked engineering samples, the same for SMP...

I investigated this quite thoroughly SMP, at the very least appears to not be routed through gold wire on tillamook PPGA and possibly not even present at all on the die. I can’t speak for the film package chips.

So ppga tillamook smp is never going to happen.

Film chip tillamook however…?

Have you noticed any curious pins on the package RayeR?

There are some Intel chips that are claimed to not be working in SMP which I have been able to explain such as the MMX overdrive I am basically 99% sure that they omitted the gold wire for one of the pins which oddly enough only disables its use as a slave CPU, so it can work as a master in SMP and I was able to do this

I am unsure how Intel locked the pentium one CPUs, but I kind of suspect that it was not through gold wire perhaps on-die fuses? I struggle to explain why some CPU examples had their multipliers scrambled. I fear we may never know unless an intel engineer shows up and and enlightens us

A note about the overdrive MMX I believe that they set its multiplier permanently through gold wire, (respective BF pins routed to vss or vcc3 accordingly) I noted that their bf pins are NC but I don’t think they did the same thing on the PPGA CPUs because that doesn’t explain how some of the multipliers got scrambled.

Something rather interesting about tillamook,

I have a bunch of the following ppga tillamooks,

233, 266

None of them are locked from 133,166,200,233,266 but I have never gotten them to do 300 at 66mhz fsb

I thought I had a 166 but I think I was mistaken, still need that!

Hm, about the pins.. If you look at the CPU back side photo that I posted on 1st page you can see that some tiny traces going out from the chip are connected together with others and then continues to solder pin as 1. I mean that there are more pins on CPU core side than there are solder pins (320). This may give a very little chance that some of them might be used to configure something - this would mean e.g. that 300MHz Tilly would have sligtly different traces patterns on the foil than others with 4x mul. If someone could take a detailed macro image of the backside then it could be compared with my photo. But I think it would be much more complicated to use different foils in manufacturing of CPU packages than if they would simply LASER trim some fuse on silicon die before sealing.

I cannot say much about SMP, just I never seen someone unlocked it.

What you mean by "scrambled multipliers"? That there are different combinations of BF pins than in datasheet table?

Here is a 266

I have encountered some p55 specimens of the same s-spec presenting at as many as 3 different mhz speeds with the same jumper settings depending on which cpu specimen was installed.

Typically this would consist of one unlocked one and two variants of locked ones. Odd behavior indeed. Which suggests something other than gold wire trickery in the multiplier locking of p55c

Good photo but 266 is also capped to 4x, right? So there shouldn't be difference. We need to get photo of 300 version.

I found an old photo I took of the T300 in that Toshiba I mentioned :

I mean photo of the back side to see routing pattern. Of course don't solder it off from your working device, maybe if someone would have a spare part or some dead laptop...

They do show up on ebay every now and then.

There is a possibility I may have one in my pile of golden warriors and golden tigers.

But I have decided not to disturb those. Without the ability to solder/repair on a microscopic scale like you, they are too risky to play with without checking them first.

Pentium 1 366mhz The Infamous Golden Tiger/Warrior

RayeR wrote on 2023-11-24, 00:00:

I would bet it's BF2:0 = 1 0 1, so let some owner try to measure it on his laptop. But I'm not sure how valuable this info would be as lower CPUs are already locked. Intel don't give bonus easily as AMD. AFAIK nobody was able to unlock multi on intel there was only some unlocked engineering samples, the same for SMP...

Thank you for the pointer image, I don't know much about the pinout on these types of chips so that helped 😀

I bought some carbide mini drills the other day and immediately broke the 0.1, 0.2 and 0.3mm bits, but there's enough left that the 0.1mm can be used as a probe tip:

That turned out to be less necessary than I thought, the BF pins are clearly routed to nearby vias:

So the layout was as you suspected, this makes it 1-0-1 right?:
BF0 vcc3
BF1 gnd
BF2 vcc3