E6400 overheating throttling

I would gladly volunteer for the testing.

Now that you mention, Tom seems to have the NVidia and T-series proc. From what I have seen so far, the issue is prevalent with the Intel GMA and P-series proc combo. Strange.

 

John,


If you put low load on the processor , like 10-15 % , do you will experience this problem despite having 30 degrees Celsius ?

 

I’ve experienced throtting problems probably going back to when I first received my system last September, though I had no way of knowing what the problem was and couldn’t figure it out back then no matter how much of my little remaining hair I pulled out. Here’s my system profile:

Dell Latitude E6500

Intel Core 2 Duo P8400 processor, 2.26Ghz

4GB RAM

NVIDIA Quadro NVS 160M graphics

Dell BIOS version A13

E/Port Plus Advanced Port Replicator (Docking Station)

dual Dell E248WFP monitors

Windows XP Service Pack 3

Note that I have the discrete NVIDIA graphics option. Some folks have suggested that NVIDIA graphics systems don’t have throttling problems. Let me emphasize (as others have) that this is simply not true.

I have studied this problem VERY closely now, have performed extensive documented testing and have created a 59-page report detailing the problem and much about the cause. The report is at:

http://imsahp.chambana.net/~randall/dell/throttlegate.pdf

In the report, I go into excruciating detail on many aspects of this problem, but here is a snippet from the conclusion:

“This article provides ample evidence to suggest that the Dell Latitude E6500 was shipped to customers with a defect that can dramatically reduce the system’s performance capacity by more than 95% under normal operating conditions, especially when docked to the E/Port Plus Advanced Port Replicator. This phenomenon has been identified in Internet forums since late 2008. Similar complaints have also been made regarding the Dell Latitude E6400. Complaints have been lodged regarding both possible graphics configurations (integrated Intel graphics and discrete NVIDIA Quadro NVS 160M graphics) for both the E6400 and E6500. A particularly unfortunate aspect of this apparent defect is that it’s unlikely it would be properly diagnosed by end users and perhaps even some technical professionals”.

Yes, you read that correctly – more than 95% performance loss, leaving less than 5% of processing capacity. Folks have complained that various utilities are reporting performance loss on the order of 65%, but what they don’t know is that there is more than one throttling mechanism at work here. In fact, there are two that are invisible to commonly used utilities (Dynamic FSB Frequency Switching and Software-controlled Clock Throttling). My report identifies a total of 10 specific throttling steps that are invoked progressively at 30 second intervals. If all 10 steps are applied, processing power is reduced to less than 5% of capacity. Again, full details are in the report, most of which consists of screen captures from testing sessions that show the throttling in action.

Some key points from the report (keep in mind this is specifically for the E6500 with the NVIDIA graphics option, but much of this likely applies to the E6400 and/or the Intel integrated graphics option):

1. The problem is NOT an overheating problem - the system simply does not overheat. It is due to premature and overly aggressive attempts at thermal control, invoked at what are NORMAL processor operating temperatures (65-80 Celsius), possibly due to faulty ACPI “passive cooling” parameter definitions and/or control methods.
2. The problem is substantially more pronounced when the system is docked.
3. The problem is aggravated somewhat by the use of dual monitors when docked as opposed to a single monitor.
4. Since the problem is all about temperature, the higher the surrounding ambient temperature in the room, the sooner and the greater the performance loss.
5. The symptoms are much more highly correlated to elevated NVIDIA GPU temperature than elevated CPU core temperatures.

Some miscellaneous corollaries:

1. Any blockage of air inlets or outlets (including, of course, dust) will aggravate the problem.
2. The reason people report shockingly high percent CPU utilization statistics when their system slows down is that the overall capacity of their processor is degraded by the throttling mechanisms. The same processes running on a CPU that is subsequently throttled necessarily will demand a higher percentage of the processor’s remaining capacity.
3. The reason some folks report persistent slowness even after installing software to prevent CPU downclocking is that more than one throttling mechanism is in play here. In particular, Software-controlled Clock Modulation (also called On-Demand Clock Modulation) occurs in an almost completely invisible manner, as opposed to performance state changes (which are usually monitored by common utilities). Another often-invisible throttling mechanism is Dynamic FSB Frequency Switching (where the FSB frequency is slashed in half), though if you prevent performance state changes, that takes care of preventing this too (since it’s part of state P3).
4. The reason there aren’t more complaints (though many are accumulating these days) is that users who experience this problem simply have no way of knowing what the cause is and are likely to blame the wrong thing (Windows, recently installed software, cooling hardware, etc.). Untold masses may be adversely affected by this problem, but nearly all of them wouldn’t know it because there’s no way for them to tell. Also, the problem is at its worst only when in a docked configuration, which may not be common.
5. The reason complaints are escalating now more than before is that this is the first summer that people have had this system (in the Northern Hemisphere, anyway). I think it’s safe to say that ambient temperatures are higher for most E6400/E6500 users now.
6. The problem can be substantially mitigated by pointing an external fan at the system.
7. The problem can also be mitigated by software, such as RMClock, that can override the throttling mechanisms in question, at the expense of negating all passive thermal management (though critical temperature shutdown mechanisms may remain in place).

A note to those who observe throttling of their system when it is essentially idle: That is Intel’s “SpeedStep” feature and it’s nothing to worry about. That only throttles the system when it is idle in order to conserve power. As soon as processor demand is present, the throttling is almost instantaneously turned off. If you’re trying to troubleshoot performance loss where throttling is sustained even when demand for processing power is high, you might want to disable SpeedStep to avoid confusion. In Windows XP, this can be done (at least on my Dell E6500 system) by choosing the “Always On” or “Home/Office Desk” Power Scheme in the “Power Options” Control Panel window.

At least one report has surfaced of an individual who has recently purchased an E6400 system with NVIDIA graphics that does NOT suffer from this defect, so new purchases may not be affected by this problem. But there may remain a large installed base of users who are adversely affected by this defect, the vast majority of whom are simply unaware of the source of their frustration. There also may remain many systems stacked in warehouses, not yet sold, that have this problem.

To date, there has apparently been no explanation or resolution of this problem from Dell.

I tried just now to make headway once again with Dell Technical Support. After explaining the problem to the Hardware Department, they forwarded me to the Software Department . I explained the problem again to the Software Department, and then they tried to forward me back to the Hardware Department 8-P. I was not too excited about that. I wound up speaking to a “second-level technical support supervisor” who said he would submit the issue to the “Escalation Team”, who would contact me directly to follow up in a day or two. He described the “Escalation Team” as very capable with a wide spectrum of expertise. Time will tell on that. Of course, I never even got the last call I was promised.

Once again, my full 59-page article on this is at:

http://imsahp.chambana.net/~randall/dell/throttlegate.pdf

A second copy is at:

http://archive.randallcotton.com/Public/throttlegate.pdf

A login/password is required for that last link – just use guest/guest

Yes, it’s a large file (25MB), sorry, but it has many large high-rez screen captures and I had to preserve their legibility, so I couldn’t compress them very much.

 

I got mine back. I have the Nvidia. They replaced the heat sink. I tested it with Call of Duty and it passed.

Oh, and they scratched the heck out of the lid. I called them and they said that I can send it back. I said "no" since I don't want to lose it for another week and it was their fault. They referred me to R something. Should I hold out for home service?

 

Your problem is when your laptop is docked only, correct?
We know this problem. The throttling problem that people faced, is the one where WITHOUT being docked, and NOT even doing anything intensive, the system throttles even after a motherboard + heat-sink replacement. That is teh problem that seams to affect Intel GPU's users only.

Hopefully BOTH problem will be fixed on the next BIOS. And I encourage everyone to push out there problem to Dell so that it reaches the people in charge in the development of the system BIOS. Just informing them, does not help, a push is needed.

 

Because it's there side, they should send you a on site technician to get it fixed. I suggest to get the Onsite service, I think you can upgarde to it if your laptop is not too old, ask Dell. Sure it cost more, but the laptop never leaves your hands. Even if Dell send you a replacement system, you hold your laptop until you get the new one, exchange the HDD (if that part is not the problem, to avoid a reinstall) and send the broken system back to them. And everything is done in front of you, and any damage done by the technician is also covered.

 

As mentioned earlier in this thread, I have the dock and the NVidia card, and I tried Tinkerdude's tests within 5 minutes after I got my machine setup and I never got any of the throttling, and this is *with* the machine docked. I let StressPrime run for close to an hour, was playing a DVD, and was doing Google Earth all at the same time and never got any throttling whatsoever. So maybe I am just lucky, but I have yet to be able to duplicate Tinkerdude's problems. Bottom line is that I'm not sure the dock is related to the throttling or not, because I have no problems at all.

 

For those of you with the Nvidia GPU, what are your idle and load temps for the chipset?

 

@dcp12345678, Ah yes yes, I completely forgot! Thanks for the reminder.
Well our friend tinkerdude is on the E6500 so maybe that explains it.

 

Asymmetricblog, Can you relay the highest CPU/GPU temps you're reaching? Did you take any measurements before you sent the unit in?

Thanks