Does anyone know exactly when they locked the clock and fsb speeds? Here, you're saying it was week 34 1998, but the Maintenance & Service Guide for Compaq Deskpro EP series (I have it on pdf) says it was 1st January 1999 for all 300mhz and faster Pentiums and Celerons.

They locked it in 1998 starting with the 333s iirc, since there was so many marked "p2 300mhz" processors that were really just p2 266mhz sold everywhere.

Overclocking a p2 did have a lot of benefit and fun from it, it's one of the only processors I know that's worth the risk of overclocking 😀 nice performance jump from 233mhz to just 290.

AdamP wrote:

Does anyone know exactly when they locked the clock and fsb speeds? Here, you're saying it was week 34 1998, but the Maintenance & Service Guide for Compaq Deskpro EP series (I have it on pdf) says it was 1st January 1999 for all 300mhz and faster Pentiums and Celerons.

1. Intel never locked the FSB speeds, the processor just indicates the intended FSB speed. Mainboards by default use this recommendation, though many retail mainboards allow you to set differing FSB speeds, like mainboards from Abit, Asus, Aopen, Soyo, MSI etc.

2. I know for a fact that the mid/late 1998 celeron 300 I have here is fully multiplier locked, so the service guide is wrong already. What is in the the second post of this topic is the best I can say with the information that I have. If you want to read the background follow the link there.

Gerwin wrote:

I intent to try a klamath, but I will be going abroad next week already... so me should either hurry or postpone it. I have this Aopen mainboard with a 1.5x multiplier option, maybe it works for klamath CPU's.

Fortunately UX-3 beat me to it, it says in german that the klamath does not respond to multiplier 1.5x:
Pentium II als DOS- & 3dfx-PC

Would voltage be an issue? Klamaths have a voltage of 0.8v higher than most Deschutes. Do Klamaths use more power than Deschutes?

http://sandpile.org/impl/p2.htm
Klamath (0.28um) is a bit more power hungry but not all that much. They're similar to a Pentium Pro actually. Compared to today's chips they are very low power processors 🤣.

If your OS or a util runs HLT instructions thru them while idling they cool down nicely too.

most sources say klamath=0.35 µ..

Apparently it was a shrink to 0.28u. P55C is on the same process. It sounds like one of those combination processes with some parts of the silicon at 0.35u and some at 0.28u. I did some digging:

http://docs.google.com/viewer?a=v&q=cache:lVI … 0bmL3mQwHQo2oKA

I decided to go ahead and test a couple of my own Deschutes P2's. The results were interesting.
I tested 5 P2's with the Deschutes core. 3 of those seemed to be completely unlocked! They would have multi's ranging from x2 all the way to x5.0 (and x5.5 in a single case even!

I'm not sure about the date codes but the 3 unlocked ones all 3 had MALAY while the 2 others had PHILIPPINES and COSTA RICA respectively.

I'll post some pictures later

Heres the pics I promised earlier
The first 2 are from an unlocked P2-400

Strangely it wouldn't post at 400Mhz with a FSB of 66, 366Mhz was the highest it would go.
The folowing 2 also were unlocked. Perhaps it might help others more easilly chase down unlocked Deschutes of their own 😉

The 2 others I tested (a 350Mhz and a 300Mhz one) were fully locked and had manufacturing locations of PHILIPPINES and COSTA RICA

All three of the unlocked ones also went down to 133Mhz btw

Enjoy!

Interesting,

So one SL2S7 P-II 400 is 'multiplier limited', and two SL2QF P-II 333's are 'multiplier limited'. (I have one SL2QF too, same thing) I doubt they are completely unlocked because they still have an upper limit for the multiplier.

I like the 400MHz most though, because it has a higher upper limit, maybe you can get 450 out of it. I will try to get me one of these later.

It is actually exactly the same observation as we made earlier, but I appreciate your additional tests: Slowing a 440BX based PC

What are the 5-digit product codes of the locked ones?

I haven't taken pics of those, the codes are SL2U3 (IRELAND) and SL2VY (PHILIPPINES), 350Mhz and 300Mhz respectively

What board are you running those on?
I threw a P II 350 (SL2U3 COSTA RICA) in an Asus P2-99 and dropped the FSB down to 66 for a nice 233MHz 😀

I'm not sure how to check if it's multiplier locked though, there are no OC/CPU setup options in the bios.

[Edit]
Totally missed the jumper block for multiplier, seems to be locked - or at least doesn't like 2x.
I'll try 2.5x and see but I somehow doubt it will change anything.

What I'm trying to do is lower power consumption as much as possible for a lightweight server.
Lowering the vcore might be the way to go from this point but the only adjustments are jumpers that are either overvolt or normal.

[Edit again]
Well, it's definitely locked at 3.5x. Ah well 😀

The SL2U3 I have is also locked.

The board I use is a chaintech 6BTM. It can set the multi in the BIOS. The board uses the venerable BX chipset and has 4 memory slots.

Seems some stepping are (semi) unlocked while other steppings are completely locked.

Time to create a list?

Seems kinda stupid that they locked them downward as well, would have been cool to see a P II at 133Mhz 😁

See 133 MHz link in my sig, TheLazy1. 😁

Neat stuff, weird how it compares to the PPro and P1 MMX.
I'm trying to go for the lowest possible power consumption here, all I need is a proper setup to measure it.
😀

Maybe they locked them downwards to prevent them from being illegally overclocked in case Intel were ever going to make a P2 with official 133Mhz FSB support?
Or maybe it was just easier to lock them completely.

I think the latter 😁

Pic:

@TheLazy1
google SL2U3 you can see that its stepping is dB0, which I would expect to be fully locked, you will have a better chance with an older stepping like dA1. example: SL2S6.

@Tetrium
The 133MHz possibility of the P-II is indeed quite neat. yet your L2 cache will not be available/working at that setting.

So I heard. Iirc the BIOS still said there was 512KB L2 cache though

The L2 problem was only concluded from speedsys results, maybe it is just speedsys at fault?