Mobile Prescott multipliers fully UNLOCKED and changeable on desktop motherboard :D

Mr Meltdown

Weaksauce
Joined
Jan 15, 2003
Messages
98
I'd been wondering if the newer Prescott Mobile CPUs could have their multilier changed in Windows as according to the spec sheets the Mobile P4 1M Prescott 533FSB CPUs should have the ability to change the multiplier using software regardless of chipset, ....

P70 of the 30242403.pdf:

6.3 Enhanced Intel SpeedStep® Technology

The processor will feature Enhanced Intel SpeedStep technology. Unlike previous implementations of Intel SpeedStep technology, this technology will enable the processor to switch between multiple voltage and operating frequency points instead of two. This will enable superior performance with optimal power savings. Switching between states will be software controlled unlike previous generation processor implementations where the GHI# pin was used to toggle between two states. Following are the key features of Enhanced Intel SpeedStep technology:

• Multiple voltage/frequency operating points provide optimal performance at the lowest power.
• Voltage/Frequency selection will be software controlled by writing to processor MSRs (Model Specific Registers) thus eliminating chipset dependency.

blah blah blah,....

Well I tried it today and Yep it works!

Abit IC7 i875 mobo / Mobile Prescott 3.06 533FSB CPU SL7DT / RightMark Clock utility: Multipliers all unlocked from 14x to 23x :D

Last picture I changed the voltage in BIOS to 1.4v instead of default 1.15V to see if the volts would stay or be lost with the clock utility software, the volts stayed allowing volts beyond the 1.35v max in the utility :D

IESS.jpg


IESS2.jpg


IESS3.jpg


IESS4.jpg


I likes this Enhanced Speedstep ALOT! (earlier Speedstep just plain sucked)

This is in my main work PC so I can't chance fuxxing up my hard disks by pushing hard just yet, I'll try it with another disk on later, but 3.6Ghz was totally rock stable @ 1.25v, the excessive volts @ 3.8Ghz was just checking the volts changed as required.

Cheers,

MM
 
But its still a presscott core, and not a Dothan or Banias. The Dothan would own the world at even 3ghz, but the prescott is just another...presscott, but its good to see the ability to fuck with the multis and not just FSB, is there '2' the nomenclature for this series? IE- 532, 542, 552, etc....and are they LGA775 aswell?
 
check benchmarks. There were some reports that I've seen before that the multis were unlocked, but they really didn't do anything. The benchmarks all seemed to show that the multiplier was always actually about 9x.
 
It's an S478 SL7DT P4 1M cache (prescott core) D0 stepping (you should have been able to work all that out from the 1st post anyway!) :). E0 would probably be better, it was just nice to see full on the fly multiplier change working in a desktop motherboard (i.e no stupid Mobile chipset requirements).

The multiplier change is real and almost instant using the RM clock utility. The BIOS is a bit mixed up though on the IC7 mobo (using latest bios) with this CPU as it see's it as a full 1Ghz or so more than it is actually running at, so it says 4.61Ghz while it's @ 3.6Ghz, but about 6Ghz @ 4.2Ghz!!.

I don't have the air cooling to hold much over 4Ghz though, plus I like some peace and quiet, I'll give it a proper try tomorrow with a different hard disk so it doesn't matter if it gets corrupted then. I want to see 17x 250 = 4250 run stable, which is maybe a bit doubtful, but 16 x 250 will be easy enough. I don't have the ram for much over 200FSB so it will be 5:4 ratio see and the mobo won't do 300FSB for 3:2 ratio (i know because I've already tried it).

CHeers,

MM
 
That's really cool because mobile P4 chips are dirt cheap on ebay.
 
Here u go:

sisoftmm.jpg

sisoftab.jpg


The adaptive mode is good it switches multiplier and voltage according to load, so 2.8Ghz (14x) light / idle, then to 3.8Ghz (19x) instantly, I like it ALOT! :D

Cheers,

MM
 
try to change the mult and run superpi on each setting.
i just had an impression (and the same intel pdf also stated) that only mobiles
beginning from 3.33ghz would have the new EIST..
cpu-z is known for false readings with mobiles..

but if the superpi runs will indeed reflect the clock-change, it's a nice find of this undocumented feature :)
 
pxc said:
All Mobile P4s (2.4GHz - 3.46GHz, NW and Prescott) have EIST: http://www.intel.com/products/processor/mobilepentium4/index.htm

But not all P4s in laptops are Mobile P4s, which could be where you got that idea. :p

i'm pretty aware of all mobile p4's having EIST.
however, only pentium M's and mobile P4s from 3.33ghz (e0 step) have MSR-controlled EIST (fast) , whereas older mobiles had chipset-controlled EIST (slow, only two multiplier settings).
this is stated at the very same pdf MrMeltdown quoted.

but seeing this kind of an EIST on 3.06ghz D0 mobile is kinda good surprise ;)
i've been looking for 3.33ghz mobile from January but i've been unable to find it.
Buying a laptop to get just cpu seems also a little weird :)
 
markku said:
i'm pretty aware of all mobile p4's having EIST.
however, only pentium M's and mobile P4s from 3.33ghz (e0 step) have MSR-controlled EIST (fast) , whereas older mobiles had chipset-controlled EIST (slow, only two multiplier settings).
It seems like all 90nm Mobile P4s have it. Pages 7 and 70 of the datasheet for the 90nm Mobile P4 confirm it's a feature of all the 90nm Mobile P4 processors, 2.8GHz-3.46GHz (518, 532, 538, 548 and 552).
 
markku said:
try to change the mult and run superpi on each setting.
i just had an impression (and the same intel pdf also stated) that only mobiles
beginning from 3.33ghz would have the new EIST..
cpu-z is known for false readings with mobiles..

but if the superpi runs will indeed reflect the clock-change, it's a nice find of this undocumented feature :)

Here u go:

30pi.jpg


38pi.jpg


Yes the higher clocked desktop Prescotts have EIST and multiplier adjustment (in BIOS usually) but have a limited selection of multipliers due to the 800 or 1066FSBs, but the Mobile 533FSB EIST CPUs will have a greater selection of multipliers available obviously :)

The Mobile Prescott is locked @ 14x in BIOS with no adjustment available, so the RM clock utility is the method and works an absolute treat being fully configurable for maximum volts and multiplier to use.

I do have a 3.2Ghz E0 now as well, but for some reason on the IC7 i have to switch HT off in BIOS to get past the POST screen!! I have no idea why that should be an issue as HT works fine on the 3.06Ghz D0. The E0 is exactly the same and does an easy 4Ghz on air with 1.4v with multipliers 14x - 24x available.

Cheers,

MM
 
Hi,

I just got 3.2 E0 (SL7NB) today but it won't post on my P4P800. Any idea???

TIA,
 
Well I've got an Asus P4P800SE and despite flashing to the latest supporting BIOS it doesn't work with any prescotts at all, which pissed me off no end I can tell you. Mind you it's a fussy one anyway, I think it suffers from "baggy socket" as it didn't work with about 1/3rd of the CPUs I've tried in it.

Make sure the socket sense pin gets good contact, I think it's still the diagonally opposite corner pin to the two key pins.

The Abits work well, but as said with my E0 Mobile 3.2 I had to switch off HT to get past the POST screen, I don't know why though.

CHeers,

MM
 
oh, ty, MrMeltdown, nice find indeed! :)

to teelek:

Unfortunately, this is known problem for asus motherboards :mad:
At some point, they stopped to support mobiles..
the earlier mobile northwoods are probably ok, but HT ones won't work.
there was a workaround, i remember, to flash with older bios,
but then again - you might lose prescott support..

if you can - try with latest bios and with some older bioses (if you have cpu that works with older bioses)
 
Thanks guys for fast response,

@ Markku,
I will try flash older bios (now 1019) thsi weekend.
any recommend which bios version I should try???

@ Mr.Meltdown
Did you try D0 on ASUS P4P800SE?
Does it work? HT?
 
teelek:

My P4P800SE didn't work with any prescott period.

The D0 worked flawlessly with HT on the Abit IC7 / IC7 Max3s (except it sees 200FSB as something like 328FSB and over states the CPU speed as a result).

The E0 Mobile I tried wont get past POST unless HT is OFF. :(

Cheers,

MM
 
Mr.MeltDown:
You may haveto do BSEL0/1 Pin Bent (200FSB) Trick in order to get IC7 work with HT.
 
I can't see how BSEL has anything to do with HT sorry, BSEL will only affect the mem dividers etc, plus it's booting @ 200 anyway in BIOS.

But feel free to enlighten me. :)

Cheers,

MM
 
teelek said:
maybe it will trick mobo that "this cpu is just desktop cpu" :D
HT has to be enabled on the chip, the FSB has nothing to do with it. for example, the first HT chip was the 3.06 with the 533 FSB.
 
I just found this thread after doing some testing with my 538 (3.2Ghz) D0 Mobile Engineering Sample that I just got on my IC7 and was wondering if anyone had worked out a way to get the BIOS report the proper FSB? I have mine set to 286 and it is showing the FSB set to 469!! :eek: CPU-Z reads it OK as 14x286 but the motherboard is still reporting stuff wrong. Also an interesting not, I cannot change any of the multipliers (14-24) in the BIOS as I had hoped since its an ES chip.
 
Back
Top