Temperature difference between CPU and its heatsink

Control: management of fans, temp/rpm monitoring via soft/hardware

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
jimmyfergus
Posts: 89
Joined: Tue Sep 07, 2004 8:24 am
Location: MA, USA

Temperature difference between CPU and its heatsink

Post by jimmyfergus » Thu Jan 27, 2005 3:07 pm

I've got a Dell, which means no onboard temperature monitoring (never again!). I recently picked a thermometer up from Best Byte (who incidentally seem cheap and fast). I've wedged the sensor between the heatsink's fins, at the bottom, and it's giving me some pretty high temps, settling on 65C on Prime94. I'm using an Acoustifan with its inline resistor and temp sensor.

So I have a question - does anyone have a feel for roughly the temperature drop between the CPU and the bottom of the heatsink fins? What do you think my real temps might be? Dell has probably used some low grade thermal interface material here, and I didn't even put any thermal paste between the probe and the heatsink!

Perhaps my CPU, a P4 2.2, is even throttling back to save itself (what's the easiest way to tell that?).

silvervarg
Posts: 1283
Joined: Wed Sep 03, 2003 1:35 am
Location: Sweden, Linkoping

Post by silvervarg » Fri Jan 28, 2005 4:13 am

The probe is probably slow at reacting and dissipates very little heat in itself. That makes the contact point between the probe and the heatsink much less sensitive to good contact, so you can do fairly accurate measurment without thermal paste there.

65C on the heatsink is really hot. Dell are known to hide away so you can't read the numbers because their "trick" to get quiet computers is to allow them to run much hotter. If they let you know how hot the CPU really gets you would complain, so they don't let you know that.

The temperature difference between the CPU chip and the heatsink fins depends on thermal paste, heatsink used and contact pressure between heatsink and CPU. So it is hard to say how much hotter the CPU is, but a good guess would be 10-20C.
I wouldn't be comfortable with running a P4 in the 85C range...

Dell might also be counting on 99% of people will not run apps that will push the maximum heat in the CPU. If less than 0.1% of people do that to the degree that the CPU burns up and they have to replace the CPU under warranty that is an acceptable price to pay for getting reputation as very quiet computers so they can sell more computers.
[size=75][b]Passive XP2500+:[/b] Convection cooled, ~XP1500+, DP-102 cooler, Abit NF7-M, CNPS6000Cu on NB, Antec Overture, Samsung P80 120GB[/size]

Ralf Hutter
SPCR Reviewer
Posts: 8636
Joined: Sat Nov 23, 2002 6:33 am
Location: Sunny SoCal

Re: Temperature difference between CPU and its heatsink

Post by Ralf Hutter » Fri Jan 28, 2005 6:10 am

jimmyfergus wrote:
Perhaps my CPU, a P4 2.2, is even throttling back to save itself (what's the easiest way to tell that?).
http://www.panopsys.com/throttlewatch.php


(BTW - I think silvervarg is 100% correct in his assesment of Dell's motives and practices., FWIW)
Main Box: Intel i3-3225, Intel DH77EB, 16GB Corsair RAM, 256GB Samsung 830, SS360GP PSU, CM PS07 case.
Music Server: Intel DH77EB + i3-3220, 2xSamsung 2TB F4, Pico PSU, Fractal Define Mini, 2xScythe Fans @250 rpm.

jimmyfergus
Posts: 89
Joined: Tue Sep 07, 2004 8:24 am
Location: MA, USA

Post by jimmyfergus » Fri Jan 28, 2005 6:44 am

silvervarg wrote:The probe is probably slow at reacting and dissipates very little heat in itself. That makes the contact point between the probe and the heatsink much less sensitive to good contact, so you can do fairly accurate measurment without thermal paste there.
Good point - I thought along those lines after I posted. I've got it wedged between the fins with a fragment of cable-tie, so the temp of the probe probaby does accurately reflect the temp of the bottom of the fin.

It's definitely an uncomfortable temperature. I think the hardest I'm likely to run the CPU is when ripping and encoding a bunch of CDs. I do develop on the machine, but a full rebuild of the codebase, which takes 20 mins, doesn't take it above 54C. Just about cool enough, assuming that means the CPU is below 75. In the summer though, the room is likely to be 5 or even 10C hotter.

So, I probably have to do something, as I don't want to put up with the Acoustifan without its inline resistor. I may go for the cheapest compromise, and just put in a Fanmate, and tolerate a bit more noise when thrashing the machine.

By the way, I would get involved in the Folding business, but since it takes my computer power consumption up by about 30-40W, I can't help feeling the money (~US$30pa + pollution) involved, might be better spent if I send it directly to worthy causes. Discuss... :). Of course, in the winter, that power is also heating the house so it's not all loss.

Post Reply