E6400 overheating throttling

Decided to watch some football today.
Windows Media Center + ATSC TV tuner. HD Channel (either 720P or 1080i, I did not check).

My system throttled most of the time. It seemed to switch between 800Mhz and 1.6Ghz. It occasionally got cool enough to unthrottle and run at 2.53Ghz.

Had Everest open and the listed temperatures looked good. However, I did not see any GMCH1 or GMCH2 temperatures listed (seems to happen when I sleep/hibernate instead of shutdown/reboot).

Anyways, the GMCH (graphics + memory controller, what Intel refers to as the "chipset") has two integrated sensors, GMCH1 and GMCH2. Does anyone know why there is such a large temperature difference between "Chipset", "GMCH1" and "GMCH2"? If they are all temperature sensors in the same chip, then shouldn't the temperature difference be small? Considering the cooling system in the E6400 is quite basic, the heat does have time to spread throughout the small chip. I guess what I'm trying to say is, I find it hard to believe that there is such a large temperature gradient in such a small chip.

 

Are you sure it was throttling, or was it the CPU adjusting its speed to suit the workload? Everest will still report the CPU speed as 1600MHz, 800MHz, etc.

You need to run a utility such as RMClock or Realtemp (see a few posts above this), which explicitly show throttling, to be sure what is happening.

John

 

I know on a Core i7 Desktop CPU when running a high stress program like Prime 95 Small FFTs, you can have a temperature gradient of 25C in a distance less than the width of a small finger nail. Some chips have a lot of current flowing through different parts of them so large temperature gradients are not unusual.

The other thing to keep in mind is that these temperature sensors are not necessarily 100% accurate. A digital number somehow implies 100% accuracy to most people but Intel has used a lot of temperature sensors that are very accurate over a very limited range but are completely inaccurate at lower temperatures. Most of the reason to monitor temperatures is so the chip can adjust itself when it starts running too hot. These sensors are usually calibrated and more accurate when reporting high temperatures compared to the temperature that they normally operate at.

I'm sure Speedstep throttling has made a lot of users think that they have a problem when it's just the way a laptop chip is designed to operate so it can be easy on your battery. If you are testing and don't want to see your multiplier sagging down then go into the Control Panel -> Power Options and set your Minimum processor state to 100% and make sure you disable C1E as well. You can do that in the bios or use a program like RealTemp to do that.

That should help keep your multiplier and MHz maxed out. Then if you see it drop at full load, you'll know that you have an actual throttling problem.

 

Watched another game last night and it was definitely throttling.
Set the power profile to high performance and after a while, it throttled to 800Mhz. Ran Prime95, no change.

 

ok so contact Dell tech support via the phone, to have your problem check out. You either don't have BIOS A19, or a faulty heatsink or motherboard.. stuff like that happens. If you have to ship your laptop because you don't have next Business day onsite service, then try to push so that they change both heat-sink and motherboard so that you don't need to send once more the system for the same problem if it's not either part that was replaced.

 

WSX: Next time your computer gets stuck at 800 MHz why not try running Prime95 and then run the tool that I posted earlier.

http://forum.notebookreview.com/showpost.php?p=5612177&postcount=435

I'd be very surprised if your laptop didn't immediately go to full power and full MHz. The Core i7 laptops are seeing up to 100% increases in their frame rates when gaming by using a similar tool that disables some of the insane throttling found in recent Dell laptops. The Latitude E6x00 series has proven throttling issues so why not try giving your computer a big shot in the arm. Make sure that EIST / SpeedStep is enabled. That can cause stuck multiplier issues.

 

@unclewebb: Thank you for MaxCore2 it's working great on Windows XP. Is there any chance to similar program for Linux?

 

I'm hoping to release a more polished replacement for MaxCore2 in the very near future with more options for Core 2 owners and it handles the i7 mobile chips as well. Easy access to the half multipliers, voltage ID and Super Low Frequency Mode (SLFM) will be nice.

Here's how it looks so far:



Unfortunately I'm not a Linux user so I can't help you. If you know any programmers that would like to make a similar app then I'll be happy to tell them what registers in the CPU need to be monitored and adjusted when the bios is trying to throttle the crap out of your laptop.

 

It looks as if you are half way to making a successor to RMClock. The other half would be nice since RMClock development stopped nearly 2 years ago.

John

 

Yeah it looks like a really nice tool!