Major CPU performance drop on Dell 9100 with P4 3.2 using battery

[Cherub]

I have a 9100 with a 3.2 P4 and an MR9800. I performed a format and reinstall with WinXP Home and I installed 1GB of Kinston PC3200.

I have found that while running on battery, performance suffers badly. I first thought that the Powerplay feature was not turning off even when deselected, but I found that it was not the case and the video clocks were not being lowered.

I have run dozens of tests with SiSoft Sandra and have isolated CPU performance drops as the culprit. Scores drop over 50% while on battery with no other changes. The CPU has no Speedstep features, and even if it did it should be running at full speed during times of high CPU demand.

I've never seen or heard of this before. Am I missing something?

 

[Abhi]

Try this...

Download SpeedSwitchXP. Set to max performance.

Did that help? If it didnt, post back.

 

[TheLonelyPhoenix]

Look through Dell's support site for possible tweaks?

 

[Cherub]

Download SpeedSwitchXP. Set to max performance.

SpeedSwitchXP is a tool for managing Speedstep features. As the P4 desktop CPUs have no such features, SpeedSwitchXP will have no effect. Regardless, I have in fact tried it and it did nothing, as expected.

Look through Dell's support site for possible tweaks?

Nothing I saw was applicable, unfortunately.

 

[jvarszegi]

I don't even understand why you want to do this. Do you understand that you'll only get half an hour of battery life if you do get this working? I thought people bought machines like yours because they didn't care about battery life at all.

 

[Cherub]

I don't even understand why you want to do this. Do you understand that you'll only get half an hour of battery life if you do get this working? I thought people bought machines like yours because they didn't care about battery life at all.

What I don't understand is why I keep getting this response. Yes, I understand that battery life will suffer. I really and truly get it. This is like telling someone that it's okay that his Porche is only producing half the horsepower it should, because it's saving gas. I'm trying to get input on what I must believe is a major issue that, as I've said, should be a deal-breaker for a very large part of the market segment for high-performance notebooks. If my system is simply defective or if the issue is correctable determines whether this notebook goes back to Dell and I have 5 days to decide.

 

[fbrdphreak]

Hmmm.....call India--I mean Dell, ask them?
Does your clock speed drop or just your "performance?" Use something like this to check if your freq drops. I had been under the impression than any Intel chip in a notebook dropped clockspeed at least some, but I might be mistaken.
Also make sure your Power Profile is set to Desktop or whatever.
Lastly, ditch that Dell and buy a real laptop.

 

[Cherub]

Does your clock speed drop or just your "performance?"

The clock never appears to drop (nor should it). I've tried CPUID's real-time clock checker, the Tom's Hardware program you suggested, and others, but I've never seen any drop in the numbers. This is also true for the video clocks. Yet, it's totally reproducable, every time. Also, I've changed the memory in the system, with no effect.

Oddly, sometimes the first one or two runs With SiSoft's CPU arithmetic test will be at full speed, only to drop down again. The SiSoft Multi-media test is even weirder: the two results bounce back and forth between full and low speed, and often conflict with each other, only to reverse positions on the next run.

This can't be a power or voltage issue, since the P4 is locked and any reduction in voltage would create instability.

I've pretty much resolved to send this back, which means no notebook for me. There's simply nothing else this powerful on the market today, and I expect that other companies' prices will be absurd when they finally get MR9800's. Even the 9100 is no longer offered, and the XPS is out of my budget.

 

[sciencewhiz]

could it be throttling the clock because of heat?

 

[Cherub]

could it be throttling the clock because of heat?

As I said above, there are no clock-throttling features on desktop P4 CPUs. Also, the problem exists regardless of CPU temps. I wish it was so easy.

 

[gryfon]

in my understanding desktop p4 does have a feature to shutdown if it's overheating, to prevent damage, maybe because cpu fan is not functioning/slowing down, maybe by unpluggin it, the cpu fan slows down ?. Maybe you can use fangui to check the temperature & the fan status of the 9100. I haven't got the chance to use it, because my i9100 is not with me now (my friend is carrying it now overseas), but i heard on notebookforums.com about this fangui, you can google it, maybe you already have it ?

regards

 

[AmigaMan]

Try Speedstep.
Why do you care if you're running on battery? Just plug it in.
j/k j/k!!!

Ok now that I got those two out of the way, have you looked at Device Manager to see if any drivers are missing? What about re-installing the version of XP that came with the machine (in your OP, I couldn't tell if you used your own copy or the OEM version). Perhaps there's something that Dell installs that limits the CPU somehow. Are there any BIOS settings you could look at. I'm just throwing out suggestions, I don't have a P4 CPU in my laptop.

 

[sciencewhiz]

Originally posted by: Cherub

could it be throttling the clock because of heat?

As I said above, there are no clock-throttling features on desktop P4 CPUs. Also, the problem exists regardless of CPU temps. I wish it was so easy.

Actually, there are. http://www.tomshardware.com/cpu/20010917/index.html

 

[Cherub]

Maybe you can use fangui to check the temperature & the fan status of the 9100

I do use it. It's a great program, but gives no clue as to what may be happening. The fans cycle as normal, the temps are in normal range all the time, and as I said above, if it's a voltage issue it would likely cause instability.

(in your OP, I couldn't tell if you used your own copy or the OEM version). Perhaps there's something that Dell installs that limits the CPU somehow

I used the WinXp disc that comes with the notebook. I was surprised that it was a real WinXp disc, as opposed to a "system restore" disc that loads a bunch or garbage onto your system.


sciencewhiz:
I suppose that there is a type of throttling that occurs with heat, but I have never understood it to be a reduction of CPU speed in the traditional sense. The temps are definitely in line; low, in fact, for a Prescott, and in any event they are certainly no higher while the computer is running on battery than when it is running on A/C.

I really appreciate all of the responses. (I was mostly ignored on notebookforums, either because the subject actually required some thought and testing or because I am one of the few who dares to use a notebook while not shackled to a wall plug.) I have, however, decided to give up the idea of a notebook, at least until more universal standards are implemented.