Uncontrollable CPU Throttling?

[Koharski]

I've got a Dell M6600 notebook (if you can call it that) with an Nvidia Quadro 4000M GPU. For months i've been trying to diagnose a problem with random lockups.

Basically, i'll be using the computer and suddenly it will become unbearably slow. Video refreshes at 1 frame per second. Hitting ctrl-alt-delete takes at least 40 seconds. It's completely unusable. Often, i'm not even doing anything strenuous on the computer. Most of the time, i'll be working on office documents with a few chrome tabs open. I formatted the computer, looked at everything that could be using all it's resources. Disabled CPU throttling in the BIOS. Nothing seemed to work.

For a month or two, it randomly seemed to get better, so I forgot about it. When it started acting up again, I decided just to live with it. The problem kept getting worse. I'd look at the task manager and see my CPU pinned at 100% usage, when all the individual processes and services added up to 20 or 30%.

I was looking in the BIOS and realized that I had CPU throttling disabled. I turned it back on. Nothing changed, but I realized that whenever my computer froze up again, the task manager would indicate that the CPU was throttling down. I couldn't figure out why, because my temps were fine. But then I realized that it was my GPU that was getting hot.

At idle, my CPU sits around 50c. GPU runs at 70 minimum. During regular use, i've seen the CPU hit 70 degrees. Using a benchmark, I managed to push it to 89 before I shut it off. But the GPU will often hit the mid 90s during standard web browsing. This seems way to high. I'm guessing that the GPU is forcing the CPU to throttle down with it? Is there anything I can do to manage this?

I don't do anything that should put a lot of stress on the GPU. I run dual monitors most of the time, and 90% of my tasks involve microsoft office and google chrome. Maybe the video ads on websites, but that's all I can think of. Maybe I could underclock the GPU somehow, or allow it to throttle down without bringing the CPU with it? I also find it strange how, with CPU throttling disabled, it will still throttle without indicating so in the task manager.

 

[LTC8K6]

Notebook CPUs tend to have a high thresh hold before they throttle.

http://ark.intel.com/products/52237...essor-Extreme-Edition-8M-Cache-up-to-3_50-GHz

Tjunction temp is 100C on that one, for example.

Have you tried to clean the dust out of the fan/cooler/vent areas of the notebook?

If the CPU/GPU are hot, you should be able to hear the cooling fan running.
Perhaps it is not running?

Could be a dust problem, too.

 

[coercitiv]

At idle, my CPU sits around 50c. GPU runs at 70 minimum. During regular use, i've seen the CPU hit 70 degrees. Using a benchmark, I managed to push it to 89 before I shut it off. But the GPU will often hit the mid 90s during standard web browsing. This seems way to high. I'm guessing that the GPU is forcing the CPU to throttle down with it? Is there anything I can do to manage this?

As LTC8K6 already suggested, first thing you should to is to make sure the GPU still gets properly ventilated: this includes making sure heat sinks are not clogged with dust and GPU fan works properly when temps rise. Removing dust from the heat sink is easy, just turn your unit off and use a vacuum cleaner near the air exhaust. Instead of touching the notebook directly with the vacuum cleaner pipe, use you hand as an interface to guide the airflow.

After making sure everything is in order on the ventilation side, you have 2-3 more options to improve system behavior:
1. Use a software like Throttlestop to disable CPU throttling when GPU temps are high (the option is available on the main screen, it's called BD PROCHOT)
2. Make an effort and repaste the GPU, but only if everything looks in order (heatsink free of dust, fan working, BIOS up to date, drivers properly installed).

What BIOS version is on the unit right now?

 

[Valantar]

On a laptop that old, you should consider disassembling it and replacing the thermal paste. I did so on my 2010 Thinkpad X201 (found the maintenance manual online, did a full disassembly, cleaned everything (the CPU with rubbing alcohol, the old thermal paste was hardened and utterly useless) and put on some Gelid GC Extreme on the CPU - done in around 30 minutes), and load temperatures sank by about 5-8C. Note that I've been regularly cleaning the heatsink and fan on this for as long as I've had it. If your fans are dirty too, the difference will be bigger. But after five or so years (the reviews on the Dell that I could find were from 2011), the thermal paste isn't going to be nearly what it was when new.

 

[Phynaz]

I'm not seeing evidence of thermal throttling. What processes take up your CPU when you are having issues?

 

[VirtualLarry]

Edit: Going off of what Phynaz wrote, and looking at other possibilities:

What is your Commit Charge shown in Task Manager when this is happening? How much physical RAM do you have?

Honestly, it sounds like it could be thrashing from excessive pagefile usage. Which can happen especially if you don't have much RAM, and have a bog-standard 5400RPM HDD.

Another possibility is some sort of rootkit or other malware, that might steal CPU cycles, and not show up in Task Manager, because that process is "stealthed" by malware / rootkit.

Edit: A failing SATA HDD / SSD can cause "pausing" too, that appears as freezes to the program you're running. But that not slow, it's actually pretty-much stopped, waiting on I/O.

I know that Chrome can be a memory hog, and if you leave it open all of the time, you could be running out of RAM and thrashing the pagefile.

 

[daxzy]

For reference, OP's computer is a SandyBridge CPU (probably 4C/8T) with a 40nm Fermi GPU (one of the hottest ever, the Quadro 4000M is rated at 100W TDP). Still, the laptop should not do what the OP is describing. Looking at this notebookcheck review (link below), they said it was actually one of the cooler running laptops they've seen (of that era, obviously), configured with an i7-2920XM and Quadro 4000M.

http://www.notebookcheck.net/Review-Dell-Precision-M6600-Notebook.65764.0.html

I think the air vents/ducts need a thorough de-dusting. Another possibility is the thermal paste has dried up and needs to be re-applied. I've done a similar task on a 2009 era laptop with a dGPU (using Arctic MX-4), and temperatures went down significantly.

If it's a malware issue, you can simply use a Linux LiveCD (e.g. Ubuntu 16.04) or Windows 8.1/10 USB boot to test that theory. If the throttling still occurs when booting to a LiveCD/USB Boot, then its likely a cooling issue.

 

[unclewebb]

I run dual monitors most of the time

When running dual monitors, the GPU is usually brought up to full speed. Try monitoring the GPU speed with GPU-Z. This creates a lot of unnecessary heat but there isn't much you can do about that. Cleaning your laptop thoroughly and replacing the thermal paste on the GPU will help. Many Dells use thermal pads. If you remove one of these, the thermal paste will not be thick enough to make up the difference. This can cause a problem where the heatsink makes poor contact with the CPU or GPU and your problems will be much worse so be careful.

Some Dell laptops also have a serious issue with excessively using Clock Modulation or Chipset Clock Modulation throttling. You can use ThrottleStop to check for that problem and correct it. Fix your GPU cooling first.