E6400 overheating throttling

Indeed. A lot of detail about "How?" is now available. The "Why?" is a mystery, but could be simple incompetence at Dell or one of their OEMs (which would reflect poorly on Dell for their lack of oversight).

Actually, Dell still has A02 available for both the E6400 and E6500 on their support site.

It's true that NVIDIA's recent defects were quite a scandal, enough to even give one pause a year later. But the Quadro NVS 160M is the newest Quadro NVS design. According to the Inquirer, the affected chips were G84, G86, G92 and G94. But wikipedia and my video driver files indicate that the Quadro NVS 160M (used as the discrete video option for both E6400 and E6500) is based on the G98 chip. (c:\windows\inf\nvdm.inf has a line that reads:

NVIDIA_G98.DEV_06EB.1 = "NVIDIA Quadro NVS 160M"

along with other similar lines that identify the chips for other Quadro GPUs. I'm running XP, by the way.)

In addition, according to the Inquirer, Dell knew about the problem in late 2007, so I think they had enough heads-up to prevent that particular defect in their E6x00 systems which started shipping a year after that.

Or not. Who knows? Maybe they're so cautious (suspicious?) about possible problems with ANY NVIDIA chip that they would go so far as to deliberately sabotage the performance of their high-end corporate docking systems as a preventative measure. But that seems kind of unlikely.

 

Also, as I explained long before, 90% of the Inquire "facts" are lies.
The Nvidia problem only reflects the G84 and G86 series on laptops only. The G9x chip provided nothing over the G8x series other than some minor performance increase, optimization, lots of bug fixes, and that problem solved. In fact Dell and other affected OEMs never hit at Nvidia on anything than the G84 and G86 series.

 

I have been able to override the throttling on my system using RMClock under XP, yes. But it's a real kludge. Using RMClock to do this does some pretty serious violence to the ACPI standard. The consequence of that could mean some things will break in unpredictable ways (maybe now, maybe later).

For instance, on my system (E6500/NVIDIA/XP), once RMClock takes over, XP's Perfmon readings of the "% of Maximum Frequency" for the CPU cores become totally unreliable because perfmon assumes that only Windows is mucking about with performance state changes. Perfmon only reflects what Windows would do regarding performance state changes (not what the actual performance state is at the time as manipulated by RMClock). That's a fairly innocuous effect by itself, but given that you're messing around with CPU frequency and voltage settings, the whole premise is really rather dicey in my opinion. Of course, geeks like us may choose to live on the edge anyway.

On top of that, RMClock:

1. Has a confusing user interface, especially to those who haven't researched Intel processor clocking features and Windows Power Schemes.

2. Could be dangerous (to the system) in the hands of the novice - for instance, with RMClock it's possible to turn off the TM1 and TM2 emergency throttling features that kick in if your CPU starts running hotter than its rated max operating temp (105C for PXX00 Core 2 Duo CPUs).

[BEGIN NEW ADDED INFO]
3. Is NOT freeware if used in a commercial/business environment.
From the license agreement: "The User may not gain any commercial benefits under the granted rights". A "payware" license would need to be purchased for each system.
[END NEW ADDED INFO]

In short, RMClock is a handy tool for geeks who are confidently knowledgable about what they're doing with it. But it's an ugly (possibly counterproductive, maybe even dangerous) solution for an average user.

In addition, I should emphasize that I can't vouch for RMClock mitigating this problem under Vista or Windows 7 (since I have no substantial experience with those yet).

You would sacrifice any mechanism that Windows might invoke to reduce your system's temperature by reducing performance. In my system at least, especially when I'm docked, that mechanism is utterly broken and serves only to sabotage performance unnecessarily. But there are situations where doing this a little bit is not a bad idea - for instance, if you'd rather deal with a somewhat slower system but have less fan noise (say, in a quiet corporate meeting) or if your system temperatures really are getting pretty dang high, but not high enough to invoke automatic emergency measures built in to the CPU or GPU.

Nope, RMClock doesn't do passive thermal management. Technically, it shouldn't even try.

I am able to with my system under XP, overriding both P-state changes and Software-controlled Clock Modulation (AKA ODCM - On-Demand Clock Modulation). But it violates ACPI, may break other things, may itself break someday (when I upgrade to Windows 7, perhaps), and it's absolutely not a practical solution, in my opinion, for the average user, especially in a workplace environment.

If we take a step back away from our own interests and consider how many thousands of people may be adversely affected by this problem without even realizing it, that brings into focus how important it is for Dell to provide a legitimate fix for what appears to be a serious defect in many (most?) of the E6X00 systems they've shipped so far. It's tempting to just install RMClock and hack things enough to get by (and if you have no other choice, I can't say I blame you), but I really would encourage folks to bang on Dell about this (I am doing so myself). It's a slippery and complicated problem, but I think that the contributions that folks have made in forums like this and the report I've put together about this issue can provide enough solid ammunition so that folks can arm themselves with enough information to make some headway with Dell so that they finally get on their horse and fix this thing.

 

RMClock seems to override the throttling if you just set up a profile with throttling set to 100%. RMClock does not try to manage temperatures but, if you do some undervolting, then you should be able to lower the maximum CPU temperatures.

Yesterday, after I disabled the throttling, the maximum ACPI temperature got up to 65C, which is only just over the maximum rated temperature for a HDD. However, it was hotter than the CPU.

See the RMClock forum for issues relating to Vista and Windows 7.

I'll try to find time to pester them next week, when I am back in UK. The effect of this problem is to give users the power of the Atom.

John

 

Thanks for all the answers. So using RMClock to override OS throttling is not recommended? I thought all that RMClock would do is somehow disable the OS from throttling...I didn't think it'd be altering voltage or frequency settings. Undervolting, which seems rather harmless, messes with voltages too, right? Would you not recommend that either?

John, has your notebook had any issues with RMClock overriding the throttle? So a safe way to override the throttle would be to do so in conjunction with an undervolt?

Sorry, I am a little unclear on this. Do you mean disabled the overriding of the throttling, or disabled the throttling? Also, is ACPI the temperature of the power supply? What does this prove? Haha, I apologize for my cluelessness on this subject.

I'm sure this has been discussed somewhere in this thread but I cannot find the answer: how widespread is this problem? Do we have an approximate bug rate among E6400 users? It seems like the severity of the bug varies from case to case...is this true?

I wholeheartedly agree. I'll definitely get on Dell's case once mine arrives.

 

Using RMClock is better than having your E6400 pretending to be a netbook. I've been undervolting using RMClock since 2004 (first started to cool down a Sony S series).

I've had no obvious issues. It is not necessary to undervolt in order to set the throttling at 100%. It is easiest to set a performance on demand profile and enable throtting with only 100% selected. Note, if you consider undervolting, that RMClock can't handle a half-multiplier (so a 2.53GHz CPU becomes 2.4GHz).

"Disabling the throttling" means overriding whatever Dell is doing. The ACPI temperature is a standard temperature sensor which is not in the CPU itself.

I suspect that many users do not give their notebook sufficient work to trigger the throttling. Today, with normal usage, my ACPI temperature is a chilly 41C.

John

 

this is a huge problem that is making a lot of people angry, check these threads out:

http://en.community.dell.com/forums/p/19247293/19513421.aspx?PageIndex=1

http://en.community.dell.com/forums/p/19270205/19519212.aspx?PageIndex=1

It doesnt affect me much unless im playing games, which i dont do often, but i am still unimpressed with the way dell is dealing with it.

 

Just to clearify, this is not a problem for everyone, atleast not for the ones with nvidia GPU. i just played some HL2 and had temps over 80C on both ACPI and CPU and over 90c for the GPU with no throttling. And the second thread is what i can see probably just a misunderstanding, a CPU throttling for a couple of seconds when starting up, or going to batteries is just normal. I still believe that this is some software/hardware issue for the intel version. maybe they have other threshold levels for "dangerous" ACPI temps?

 

Acidspy,

I notice you post this message all over this forum, but this information contradicts other reports. Check the other posts in this thread. What makes you so convinced?

 

I'm not sure what you mean by "all over this forum", i have just posted this here and mentioned the heat issue in another thread. And I dont think it contradicts other reports, the throttling issue is first (and mostly) confirmed with those who have the Intel version. One other important thing is that I dont have the problem, so it cant be affecting eveyone. Dont you think that is good to know?