PDA

View Full Version : P4 frequency halved on batteries... I WANT THE REAL STORY...


ozzlo
04-04-05, 02:16 PM
ok for as long as I can remember the p4 has always had it's speed halved on batteries but every now and then I hear of a time where it is not... :shrug:

I just want to bust this myth and turn it into fact or fiction...


So what I want to know is....


is this effect permanent?

can the setting be changed in bios or windows?

If the setting is modified with windows power profiles... which power profile sets what state for the processor?

what happens to the battery life if it actually dosen't halve the frequency???

core voltage drop?

What type(s) of p4's does this effect (pentium 4-m, mobile pentium-4, mobile celeron, mobile prescott p4)?

what about the bus speed?


I have a theory in the works that the pentium4-m had the permanememt setting but with the future p4's the setting can be changed with the windows power profiles somehow I will probbably figure out if my theory is correct or not within a few replies to this thread...




the person or persons that can get me the most useful info will get credit in my mobile chip guide stickey :santa:

Quailane
04-04-05, 02:19 PM
I don't think this is true on my dad's 3.4Ghz dell XPS laptop. Even on batteries it still shows up as 3.4Ghz.

spyrollinc
04-04-05, 02:27 PM
My A75-S211 has a 538 in it. Mobile P4 with HT. When on battery power it drops the multi down to 14 (24 plugged in) and drops the core down to 1.116 (1.31 plugged in) according to CPUz. I can change the state using Toshiba's power management program. I can select Low/Mid/High for processor speed, but low and mid seems to keep the multi at 14 still. I'll play around with it some more later.

ozzlo
04-04-05, 02:43 PM
ah yes... I forgot that jigpu had told me his drops to 2/3... but still I need more data as this is one thing that I have overlooked for years and always assumed....

"assuming is the source of all **** ups" I think that came from the movie speed 2 or some similar movie involving a large boat and some suicidal maniac that decides to take it over...

Lunatic_Freak
04-04-05, 02:59 PM
My Dell M70 Pentium M 750 run at 1.86Ghz (14x multi) on AC and about 798Mhz (6x multi) on battery.
I can force it to run perminently at 1.86Ghz by changing my power scheme to "Always On" in the windows XP power settings.

macklin01
04-04-05, 03:32 PM
It depends if it's a desktop Pentium 4 or a mobile pentium 4, or it's earlier cousing, the Pentium 4-M. The latter two have speedstep technology, and the former does not. So a laptop with a regular P4, such as the Dell XPS, will not drop in speed, but a laptop with a mobile chip might, depending upon its configuration.

Also, be sure to not include the P4-M and mobile P4 with the Pentium M and Celeron M. The latter two are a whole different architecture that don't perform quite as well as a P4 (but do work admirably well), but have tremendously better power and heat characteristics. -- Paul

ozzlo
04-04-05, 04:36 PM
The latter two are a whole different architecture that don't perform quite as well as a P4 (but do work admirably well), but have tremendously better power and heat characteristics. -- Paul


BAH! don't perform quite as well... The high end p4's that are in laptops start throttleing easily because the cooling on them can't handle the heat output... I mean seriously the xps gen1 had an EE chip in it 100+ watts, but like EE's require major cooling or they will throttle...


plus on batteries the pentium-m will own the p4 if the p4 throttles down the clock. and if the p4 dosen't throttle down the clock the battery on it ain't gonna be able to last to render more than 45 mins...

macklin01
04-04-05, 05:44 PM
Well, yeah. Clearly the P4 isn't ideal for laptop use. Why do you think I've been encouraging the PM architecture over the P4 arch for desktop use? Architecturally, the P4 is a more powerful chip. Period. But it's definitely not the ideal chip for a laptop given its heat an power characteristics, as you said.

spyrollinc
04-04-05, 11:07 PM
I dont think the P4 is that great. Clock for clock the PM is faster. I think the shorter pipeline on the PM hurts its ability to get higher clocks, but its not always important. look at the amd chips and its PR rating.

macklin01
04-04-05, 11:21 PM
Certainly. I'd absolutely agree that the PM is a more efficient, largely superior design. You get a lot of computational bang for the wattage buck. I'm excited to be getting one to work with this week. But the P4 still is computationally more powerful. I don't mean in terms of clockspeed, but in terms of bona fide computational power. There are applications where raw computational power are more important (CFD, AutoCAD, gaming), and there are applications where efficiency are more important (laptops, silent computing, perhaps even general desktop use). There's no value judgement to be made here: the P4 is computationally more powerful when compared to the PM, and the PM is a superior choice for mobile and small form factor applications. Both have their strengths and weaknesses. It's not a battle of the architecture camps here. :) Just an objective comparison of the relative computational power and heat/power characteristics of two types of chips. -- Paul

Lunatic_Freak
04-04-05, 11:51 PM
There are applications where raw computational power are more important (CFD, AutoCAD, gaming),

Dont count out the Pentium M for 3D CAD work so quick, I bought my Dell M70 specfically for 3D CAD work. Take a look at this link from AutoDesk (http://www.inventor-certified.com/graphics/SV10_Puiblic_by_time.pdf), its organized by the total time to complete the benchmarks (forth row in from the right). AutoDesk actually just posted this info today.


Notice how the good old Pentium M 1.86gGhz Dell M70 is right there on the first page surounded by 3ghz+ workstation class desktops.

macklin01
04-05-05, 01:19 AM
That's very interesting, and thank you for the link. It's actually much closer than I expected. They hold up very well. I wish there were more testing on the impact of the graphics card on the PM machines. I noticed that for a fixed processor and machine, there was a decent amount of variation due to the graphics card used. (Makes sense: the software should be written to take full advantage of the higher end graphics cards.)

It's too bad they had numerous typos, though. There is no such thing as a P4-M at 1.86 GHz, but there is a PM at 1.86 GHz. (Dell M20, M70 workstations) A look at Dell's website confirms that it's a PM 1.86 GHz, not a P4-M.

It would also be interesting if they gave info not only on how much memory was used, but what type. The PM's are all with 533 MHz bus (which is evidently quite helpful--fascinating!) and DDR2 memory in a dual channel configuration. Also interesting that some of the 64-bit chips were made to run with a 32-bit OS and others were not. From the looks of this, memory performance is important to the application, and as I recall, Xeon's typically don't use DDR2 and have a slower bus speed.

Very interesting information. It shows that a top-of-the-line PM performs right up there near the top-of-the-line P4's and Xeon's. Thanks again for the post! -- Paul

spyrollinc
04-05-05, 04:10 AM
ok well lets get back to the subject at hand. multis, fsb, voltages on/off battery and so forth.

dustybyrd
04-05-05, 04:16 AM
Also, be sure to not include the P4-M and mobile P4 with the Pentium M and Celeron M. The latter two are a whole different architecture that don't perform quite as well as a P4 Paul


i agree that in some benchmarks (maybe most) the P4 3.2ghz in the laptop is faster than a 2ghz Pentium M...

but, there are benchmarks where the 2ghz Pentium M beats the 3.2ghz P4 Laptop (or desktop) in raw computational power...

for example, in superpi 1M the P4 3.4 ghz desktop scores about 41-43s...my Pentium M 2ghz laptop does it in 41s...so this must be faster than a slower fsb, 3.2-3.4ghz P4 laptop...

also in Lame MP3 encoding, the 2ghz Pentium M laptop beats the 3.2ghz P4 laptop:

http://www.gamepc.com/labs/view_content.asp?id=gmso&page=8


but, more importantly, i have now been able to overclock the 2ghz Pentium M in my Dell 600m to 2.4ghz stable (120mhz FSB)...

did superpi 1M in 34s (no patch)...

pcmark2002=8100 (my p4 3.4ghz desktop scored 8500, iirc, with 288mhz FSB)...

pcmark 2004=4800 (my p4 3.4ghz scored 4900---and that's WITH hyperthreading, that helps quite a bit in pcmark2004)...

dark_15
04-05-05, 08:01 AM
I don't know if this is much of help, but my lappy (which has a PIII tualatin) drops from 1 GHZ to 731 MHZ (Multi drops from 7.5 to 5.5). When the battery is critically low (10%-1%), it will drop to 500 mhz.

If I plug it in with more than 50% of the battery life left, it will move up to 1GHZ again. If it's between 10% to 50%, it will stay downclocked @ 731mhz until I restart the machine. If it's less than 10% it will revert back to 731mhz... not 500 mhz.

Don't ask me why... :shrug:

Slackfumasta
04-05-05, 09:27 AM
With my 5160, which has a P4-M 3.06 GHz processor, it typically runs at 1.83 GHz with or without power until I do something that needs the processing speed, at which point it jumps up to full speed whether it's on battery or not. I've only watched it with Mobilemeter, but maybe I'll watch it with CPU-z and double check what's going on.

ozzlo
04-05-05, 03:58 PM
I was getting all excited when I logged on and saw 15 posts... I was like expecting my question to have a very nice solid answer but instead i found a 10 post long thread hijack... LOL

I need to know the specifics like WHICH type of p4 manages it's frequency in what way...


that was some nice info on the p3... I diden't know that the p3 downclocked as well but now that I think about it the speedstep from the p4 came from the p3...

macklin01
04-05-05, 05:07 PM
It depends if it's a desktop Pentium 4 or a mobile pentium 4, or it's earlier cousing, the Pentium 4-M. The latter two have speedstep technology, and the former does not. So a laptop with a regular P4, such as the Dell XPS, will not drop in speed, but a laptop with a mobile chip might, depending upon its configuration.

Answer is right there. ;) And of course the Pentium M also has speedstep. -- Paul

ozzlo
04-05-05, 06:24 PM
what about the vcore, fsb and stuff like that... I am not convinced that the desktop p4 diden't drop down on batteries... I am convinced that all of them drop frequency in one way or another... sorry for beating this to death but I don't want to get it wrong... I will look into this too...

macklin01
04-05-05, 06:53 PM
Try here:
http://www.intel.com/products/processor/index.htm?iid=HPAGE+header_products_processors&

And more specifically, here:
http://www.intel.com/products/processor/pentium4/index.htm

The only desktop P4's with any kind of speedstep are the newest 6xx chips. (Undoubtably there to deal with the terrible heat and power issues of the Prescott cores) Here's another link:

http://www.overclockers.com.au/article.php?id=357950

I hope this proves useful for you. -- Paul

*edit* Might you be confusing speedstep with the thermal throttling safeguards of all P4's? */edit*

kevin_bouchard
04-05-05, 07:35 PM
Myself, I just bought a 2.2ghz P4 model D220S made by clevo. The defaults in the bios are set to throttle 50%, making the speed 1.1ghz on the battery. Luckily this is changeable and I was able to disable it. Then I used RightMark CPU clock utility to throttle it when I dont need the extra speed. Keeps the laptop cool and quiet. The only problem I've found is that rightmark detects that the cpu is throttling(if set through the bios or through the program) but the clock doesnt actually change; it just goes into "throttling mode". It does however increase my battery times and temps while reducing performance when I lock it at a lower clock so it must be working.
I do remember toms hardware doing a review on a dell that did clock down to 1.1ghz on batteries(advertised as 2.2ghz) and there was no way of changing it. So it seems that it is model dependent.
Oh and my cpu is a Northwood.

Hope this helps out.

ozzlo
04-06-05, 02:48 PM
Try here:
http://www.intel.com/products/processor/index.htm?iid=HPAGE+header_products_processors&

And more specifically, here:
http://www.intel.com/products/processor/pentium4/index.htm

The only desktop P4's with any kind of speedstep are the newest 6xx chips. (Undoubtably there to deal with the terrible heat and power issues of the Prescott cores) Here's another link:

http://www.overclockers.com.au/article.php?id=357950

I hope this proves useful for you. -- Paul

*edit* Might you be confusing speedstep with the thermal throttling safeguards of all P4's? */edit*


those are desktop chips... not notebook chips :rolleyes: I'm talking notebook chips specifically the pentium 4-m and the mobile pentium 4 (no HT involved... I know that the newer chips you can kindof use power profiles for your running speed but the effects of running on batteries for the old versions (the first two generations of the p4 400mhz & 533mhz fsb versions) is specifically what i'm asking about...


no I'm not confusing it with thermal throttling...

Slackfumasta
04-06-05, 02:50 PM
What do you mean by no HT involved? I have a P4-m with HT. . .

ozzlo
04-06-05, 04:03 PM
according to intel... that's not a p4-m... it's a mobile p4...


ALL OF THEESE CHIPS ARE EVERY VERSION OF THE P4-"m" ever made!!!!!!!!!!!!!!!!!!!
sSpec# CPU Speed Bus Speed Mfg. Tech Stepping Cache Size Package Type
SL6WY 2.50 GHz 400 MHz 0.13 micron D1 512 KB 478 pin PPGA FC-PGA2
SL6P2 2.50 GHz 400 MHz 0.13 micron C1 512 KB 478 pin PPGA FC-PGA2
SL6VC 2.40 GHz 400 MHz 0.13 micron D1 512 KB 478 pin PPGA FC-PGA2
SL6K5 2.40 GHz 400 MHz 0.13 micron C1 512 KB 478 pin PPGA FC-PGA2
SL6LS 2.40 GHz 400 MHz 0.13 micron C1 512 KB 478 pin PPGA FC-PGA2
SL6J5 2.20 GHz 400 MHz 0.13 micron C1 512 KB 478 pin PPGA FC-PGA2
SL6LR 2.20 GHz 400 MHz 0.13 micron C1 512 KB 478 pin PPGA FC-PGA2
SL6VB 2.20 GHz 400 MHz 0.13 micron D1 512 KB 478 pin PPGA FC-PGA2
SL6CL 2 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL6DF 2 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL6FK 2 GHz 400 MHz 0.13 micron C1 512 KB 478 pin PPGA FC-PGA2
SL6V9 2 GHz 400 MHz 0.13 micron D1 512 KB 478 pin PPGA FC-PGA2
SL6CK 1.90 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL6DE 1.90 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL6V8 1.90 GHz 400 MHz 0.13 micron D1 512 KB 478 pin PPGA FC-PGA2
SL6FJ 1.90 GHz 400 MHz 0.13 micron C1 512 KB 478 pin PPGA FC-PGA2
SL6FH 1.80 GHz 400 MHz 0.13 micron C1 512 KB 478 pin PPGA FC-PGA2
SL6CJ 1.80 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL65Q 1.80 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL69D 1.80 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL6V7 1.80 GHz 400 MHz 0.13 micron D1 512 KB 478 pin PPGA FC-PGA2
SL6FG 1.70 GHz 400 MHz 0.13 micron C1 512 KB 478 pin PPGA FC-PGA2
SL5ZZ 1.70 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL5Z7 1.70 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL6CH 1.70 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL6V6 1.70 GHz 400 MHz 0.13 micron D1 512 KB 478 pin PPGA FC-PGA2
SL5ZY 1.60 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL6CG 1.60 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL6FF 1.60 GHz 400 MHz 0.13 micron C1 512 KB 603 pin PPGA FC-PGA2
SL5YU 1.60 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL6CF 1.50 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL5ZX 1.50 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL5YT 1.50 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL5ZH 1.40 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL5ZW 1.40 GHz 400 MHz 0.13 micron B0 512 KB 478 pin PPGA FC-PGA2
SL6NA 1.30 GHz 400 MHz 0.13 micron B1 1 MB 479 pin H-PBGA
SL6P4 1.10 GHz 400 MHz 0.13 micron B1 1 MB 479 pin H-PBGA FC-BGA2

i'm gonna add that to my stickey I'm working on...


Theese are "MOBILE" pentium 4 processors

I think only theese that are over 3ghz had hyperthreading...

sSpec# CPU Speed Bus Speed Mfg. Tech Stepping Cache Size Package Type
SL7X5 3.33 GHz 533 MHz 90 nm E0 1 MB 478 pin PPGA
SL7DU 3.20 GHz 533 MHz 90 nm D0 1 MB 478 pin PPGA
SL77R 3.20 GHz 533 MHz 0.13 micron D1 512 KB 478 pin PPGA
SL726 3.06 GHz 533 MHz 0.13 micron D1 512 KB 478 pin PPGA
SL7DT 3.06 GHz 533 MHz 90 nm D0 1 MB 478 pin PPGA
SL77P 3.06 GHz 533 MHz 0.13 micron D1 512 KB 478 pin PPGA
SL77N 2.80 GHz 533 MHz 0.13 micron D1 512 KB 478 pin PPGA
SL725 2.80 GHz 533 MHz 0.13 micron D1 512 KB 478 pin PPGA
SL7DS 2.80 GHz 533 MHz 90 nm D0 1 MB 478 pin PPGA
SL77M 2.66 GHz 533 MHz 0.13 micron D1 512 KB 478 pin PPGA
SL724 2.66 GHz 533 MHz 0.13 micron D1 512 KB 478 pin PPGA
SL723 2.40 GHz 533 MHz 0.13 micron D1 512 KB 478 pin PPGA

and now for "MOBILE" prescotts... oh darn they don't have em....

but they have this instead...

http://www.intel.com/products/processor/mobilepentium4/index.htm


It seems that HT starts on 2.66ghz but there is also an exact similar line starting at 2.4ghz that don't have HT...


anyways... back to the the thread topic and how being on batteries affects the p4...

spyrollinc
04-06-05, 09:14 PM
i have a mobile pentium 4 with HT. its a prescott too. dunno why you said they dont have em. i can post a screen if you want.

ozzlo
04-07-05, 10:55 AM
They don't have detailed spec sheet for em...

spyrollinc
04-07-05, 03:23 PM
heres my cpuz screenie.

Slackfumasta
04-07-05, 04:44 PM
That's what I meant. I have a mobile P4 HT, Prescott core.