CPU temp jumps when I turn on external HDD

ginahoy · Post by **ginahoy** » Sat Aug 02, 2008 11:00 am

I recently installed CoreTemp and SpeedFan. When my system is idle, my CPU is stable at about 34C to 35C. I noticed when I turn on the external HDD, the CPU temp jumps 10C and remains there as long as external drive is on. During this time, CoreTemp reports the CPU is operating at 0% to 1% load.

So if the CPU is idle, what could be causing this temperature spike? Do others see this?

Lawrence Lee · Post by **Lawrence Lee** » Sat Aug 02, 2008 11:14 am

Not sure why its happening but that exact same thing happened to me a couple of years ago.

Plugging in a USB drive (key or enclosure) caused the CPU temp to spike 5C immediately. Some time later my USB-connected remote control started to behave erratically and then my southbridge chip burned out and my system was toast.

Matija · Post by **Matija** » Sat Aug 02, 2008 11:33 am

That doesn't look too good...

Check what happens to CPU voltage. Maybe there's something wrong with the motherboard, and something might fry.

ginahoy · Post by **ginahoy** » Sat Aug 02, 2008 12:20 pm

Matija wrote:That doesn't look too good...

Check what happens to CPU voltage. Maybe there's something wrong with the motherboard, and something might fry.

I logged Vcore in SpeedFan and when no USB is connected, Vcore is 1.4V. When I turn on the ext HDD, it drops about 1% to 1.385V. This behavior is repeatable, but I suspect its at the edge of precision for the hardware monitoring chip.

BTW, I confirmed plugging in a SanDisk also causes CPU to spike 10C.

SpeedFan also reports a spike in "Temp 3" but I haven't been able to confirm what this sensor refers to on my mobo. Someone said it may be the SouthBridge temp.

You guys have me worried now! I built this system for AutoCAD and possibly 3D renderings but haven't installed those apps yet. I'd like to get this resolved before I start pushing the CPU.

What's a good discussion forum for overclockers? Since my mobo is (was) popular with overclockers, I was thinking those folks might know what's going on.

Trekmeister · Post by **Trekmeister** » Sat Aug 02, 2008 4:45 pm

There is/was a problem with some USB devices preventing the CPU from reaching the lower idle states. I believe it was mostly noticeable on laptops since a plugged in USB device reduced battery time more than it really should. Perhaps your increasing CPU temps could be a symptom of the same problem?

dhanson865 · Post by **dhanson865** » Sat Aug 02, 2008 5:11 pm

Trekmeister wrote:There is/was a problem with some USB devices preventing the CPU from reaching the lower idle states. I believe it was mostly noticeable on laptops since a plugged in USB device reduced battery time more than it really should. Perhaps your increasing CPU temps could be a symptom of the same problem?

Faulty Microsoft Driver Saps Intel Core Duo power
http://it.slashdot.org/comments.pl?sid= ... d=14588517

though if the driver did so for one chip it may do so for others. It's a Microsoft bug not an Intel bug.

ginahoy · Post by **ginahoy** » Sat Aug 02, 2008 10:26 pm

dhanson865 wrote:Faulty Microsoft Driver Saps Intel Core Duo power
http://it.slashdot.org/comments.pl?sid= ... d=14588517

though if the driver did so for one chip it may do so for others. It's a Microsoft bug not an Intel bug.

Wow! You guys are awsome! This was indeed this issue. The registry tweak fixed the problem!

Gaining 10 degrees C of additional headroom is HUGE!!! Every Silent PC enthusiast should be aware of this issue and verify their system isn't affected.

Interestingly, although the USB/EnIdleEnpointSupport key wasn't in the CurrentControlSet folder, I found it in the ControlSet003 folder. I sure didn't put it there. Unfortunately, I don't know enough about the registry to understand why there are several ControlSet folders.

Trekmeister · Post by **Trekmeister** » Sun Aug 03, 2008 6:19 am

ginahoy wrote:Wow! You guys are awsome! This was indeed this issue. The registry tweak fixed the problem!

Gaining 10 degrees C of additional headroom is HUGE!!! Every Silent PC enthusiast should be aware of this issue and verify their system isn't affected.

Glad to be of help!

Still, this only affects your idle temps. It should not give you any more headroom under load since this was an idle problem only.
Nonetheless, the saved power when idle might give you a more quiet machine and a slightly lower power bill.

ginahoy · Post by **ginahoy** » Sun Aug 03, 2008 9:20 am

Trekmeister wrote:Still, this only affects your idle temps. It should not give you any more headroom under load since this was an idle problem only.

Yes, sorry. I should have been more specific -- I'm talking about headroom at low-load conditions. To say that the problem only effects idle temps is not accurate. I'm not sure how many power states my CPU has, but the USB issue apparently leapfrogged more than one state.

During a typical day of browsing and messaging, my CPU bumps along at 32C to 39C (SpeedFan provides a neat graph). I've never seen it increase by 10C, as it did with a USB device. So while it's true that fixing this issue doesn't provide additional headroom at full load, it benefits all other load conditions, which in my case, is 100% of the time.

Trekmeister · Post by **Trekmeister** » Sun Aug 03, 2008 1:28 pm

ginahoy wrote:So while it's true that fixing this issue doesn't provide additional headroom at full load, it benefits all other load conditions, which in my case, is 100% of the time.

Excellent point. Only looking at the full load case (as I did) is a too simplistic view. Real life is usually a bit more varied.

dhanson865 · Post by **dhanson865** » Sun Aug 03, 2008 6:33 pm

ginahoy wrote:I'm talking about headroom at low-load conditions. To say that the problem only effects idle temps is not accurate. I'm not sure how many power states my CPU has, but the USB issue apparently leapfrogged more than one state.

During a typical day of browsing and messaging, my CPU bumps along at 32C to 39C (SpeedFan provides a neat graph). I've never seen it increase by 10C, as it did with a USB device. So while it's true that fixing this issue doesn't provide additional headroom at full load, it benefits all other load conditions, which in my case, is 100% of the time.

Glad to know you see such noticeable gains.

As CPUs get faster and faster we will spend much more time at idle and/or some mid power state. A bug like this is likely to be affecting a large number of users.