Core Temperature Question

Message boards : Number crunching : Core Temperature Question
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Allie in Vancouver
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 3949
Credit: 1,604,668
RAC: 0
Canada
Message 1056442 - Posted: 15 Dec 2010, 23:10:51 UTC

I'm just a little curious about the wide difference in core temperatures in my Q6600. When crunching under full load cores 0 and 1 have temperatures around 60-62C but cores 2 and 3 run about 10 degrees cooler, around 50 -52C.

It is not a huge issue since it has always been that way, but I am curious.
Pure mathematics is, in its way, the poetry of logical ideas.

Albert Einstein
ID: 1056442 · Report as offensive
Profile SciManStev Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Jun 99
Posts: 6651
Credit: 121,090,076
RAC: 0
United States
Message 1056445 - Posted: 15 Dec 2010, 23:22:05 UTC - in response to Message 1056442.  

I'm just a little curious about the wide difference in core temperatures in my Q6600. When crunching under full load cores 0 and 1 have temperatures around 60-62C but cores 2 and 3 run about 10 degrees cooler, around 50 -52C.

It is not a huge issue since it has always been that way, but I am curious.


I'm going to guess at this one. Perhaps the cooling sink is not perfectly distributed. My cores can vary a bit too, and sometimes up to 10°C like yours. With hyperthreading on, the different threads of the same core are very close in temperature. These are my temps.



Steve
Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website
ID: 1056445 · Report as offensive
archae86

Send message
Joined: 31 Aug 99
Posts: 909
Credit: 1,582,816
RAC: 0
United States
Message 1056455 - Posted: 16 Dec 2010, 0:07:59 UTC - in response to Message 1056442.  

I'm just a little curious about the wide difference in core temperatures in my Q6600.

If it is real, that sounds bad, but it may well not be real. The on-die therometers are bad in a number of ways, and on-die matching is one of them.

If you take the system to idle, does the matching get better? If the problem is real, it should, as the mismatch should in some sense multiply a thermal impedance by a power flow, and with greatly reduced power flow, the difference should decrease.

Sadly some of the sensors are also non-linear and even "sticky" across temperature range, so my suggested sanity check is far from definitive.

Intel meant these things as part of a scheme to get adequate population reliability by holding down most cases of extreme over-temperature (assuming something in the system responded to high reported temperatures by active control--whether of cooling, clock rate...) They are not claimed to be thermometers, and they actually are not very good ones.
ID: 1056455 · Report as offensive
Dena Wiltsie
Volunteer tester

Send message
Joined: 19 Apr 01
Posts: 1628
Credit: 24,230,968
RAC: 26
United States
Message 1056459 - Posted: 16 Dec 2010, 0:23:32 UTC

Even IBM has problems with that. My Power PC G5 Mac runs 162f on one core and 155f on the other core under load. Don't worry, that is in the safe range for the chip.
ID: 1056459 · Report as offensive
Profile SciManStev Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Jun 99
Posts: 6651
Credit: 121,090,076
RAC: 0
United States
Message 1056460 - Posted: 16 Dec 2010, 0:24:16 UTC - in response to Message 1056455.  
Last modified: 16 Dec 2010, 0:26:32 UTC

That was a very excellent and knowledgeable reply. You made me curious what my idle temps are.



Steve
Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website
ID: 1056460 · Report as offensive
Profile Allie in Vancouver
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 3949
Credit: 1,604,668
RAC: 0
Canada
Message 1056463 - Posted: 16 Dec 2010, 0:50:23 UTC

At idle the discrepency is about 5 degrees C (37C for cores 0 and 1, 32C for cores 2 and 3)

I am fairly sure that it isn't a really serious issue because it has been that way since I first fired this computer up in October 2008. If it is a fatal problem I imagine it would have sunk me by now.
Pure mathematics is, in its way, the poetry of logical ideas.

Albert Einstein
ID: 1056463 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 34744
Credit: 261,360,520
RAC: 489
Australia
Message 1056465 - Posted: 16 Dec 2010, 0:56:48 UTC - in response to Message 1056442.  
Last modified: 16 Dec 2010, 0:59:22 UTC

I'm just a little curious about the wide difference in core temperatures in my Q6600. When crunching under full load cores 0 and 1 have temperatures around 60-62C but cores 2 and 3 run about 10 degrees cooler, around 50 -52C.

It is not a huge issue since it has always been that way, but I am curious.

Remember that these are made up of 2 dual core processors and its very hard to get 2 dies that will read the same temps, plus some motherboard sockets orientate the package in such a way that 2 of the cores sit above the other 2 (this is likely in your case) instead having them all sit in a horizontal line (heat does rise making the upper cores warmer) so its really nothing to worry about unless you start getting around the 70C mark though at around 72C the cores will start throttling themselves down or completely shut down if the temp still rises.

Cheers.
ID: 1056465 · Report as offensive
-BeNt-
Avatar

Send message
Joined: 17 Oct 99
Posts: 1234
Credit: 10,116,112
RAC: 0
United States
Message 1056517 - Posted: 16 Dec 2010, 5:03:55 UTC
Last modified: 16 Dec 2010, 5:04:37 UTC

Please keep in mind Intel has stated that the on die temp sensors are not accurate below 60 c. So variances can be artificial. The best way to test them is heat them up and see where they lay, a 5 degree temp difference is pretty normal due to the complexities of the core and micro-architecture. 10 degree difference can be due to a heat sink mounting issue, more than likely the spread or lack there-of of thermal paste. Anything more than that is a stuck sensor.

As long as you are staying below certain thresholds. You have two different step type on the q6600 B3 and G0. The B3's max temp is 62.2 and the G0's is 71. Those are the "melt down" temps you should be fine as long as you stay under 60 long term. Which you appear to be, so you're good to go, any more voltage or clock and you will need a better cooling solution.
Traveling through space at ~67,000mph!
ID: 1056517 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 19012
Credit: 40,757,560
RAC: 67
United Kingdom
Message 1056567 - Posted: 16 Dec 2010, 7:46:08 UTC

My Q6600, G0, temps are
100%	0%
47	39
46	38
39	33
39	33


Which also indicates I should have done some cleaning when the project was down. Just after cleaning of h/sink and fans etc the high temp is usually ~43C
ID: 1056567 · Report as offensive
DJStarfox

Send message
Joined: 23 May 01
Posts: 1066
Credit: 1,226,053
RAC: 2
United States
Message 1056624 - Posted: 16 Dec 2010, 15:33:21 UTC

I have a Dell computer at work that is a Q6600. My idle core temps are within 3C and my load core temps are within 6C. Seems to be a semi-normal thing with this chip.
ID: 1056624 · Report as offensive
Profile Fred J. Verster
Volunteer tester
Avatar

Send message
Joined: 21 Apr 04
Posts: 3252
Credit: 31,903,643
RAC: 0
Netherlands
Message 1057272 - Posted: 17 Dec 2010, 23:29:02 UTC - in response to Message 1056624.  
Last modified: 17 Dec 2010, 23:40:23 UTC

I've build 2 computers, used ASUS P5E, X38 chipset.
One has 1 GTX470 and 480, but only run in PCI-E x2 (470) and 1x for 480.

SETI MB's appear to have little trafic on this bus, cause I notice no difference and run 2 MB WU's a 1 card, now f.i. 4x Einstein on CPU and 4 SETI CUDA FERMI, on the GPU's
Only at GPUgrid, I can notice the difference in speed increase.

That same Mobo, works with 1 HD4850 and 1 HD5870 and both run in PCI-E x16. ver2.0, according to GPUz 0.49 and is also the Q6600 that runs sometimes @ 103C,
according to the program, that came with the mobo. That host runs 2 years non stop, 24 x7 x365 (-0.1%), it seldom throttles back, the mobo temps are also quite high: 68C, due to the 4850 & 5870. (This host uses more then FERMI, 530Watt, FERMI 460 Watt, 100% crunching on CPU and GPU's.

The ATI host has, when crunching a very heavy screen-lag, takes a minute, maybe less, before a program shows his 'face'. You can follow the build up of the screen! Impossible to do anything else on that host, it does MW now.
Unlike the (XP64) GTX470 & 480 host, on which you hardly notice, that the video card is also doing something else. (ATI host run WIN XP x86, FERMI host runs XP64.

I think, I have to get another mobo!
ID: 1057272 · Report as offensive
Profile Sutaru Tsureku
Volunteer tester

Send message
Joined: 6 Apr 07
Posts: 7105
Credit: 147,663,825
RAC: 5
Germany
Message 1057278 - Posted: 17 Dec 2010, 23:46:02 UTC
Last modified: 17 Dec 2010, 23:47:37 UTC

With my Intel Core2 Extreme QX6700 (2.66) @ 3.14 GHz (Quad-Core-CPU) it was normal up to ~ 8 °C temp differences (min. to max. Core).

The Q(X)6xxx are two Duo-CPUs.

Currently with my Intel Core2 Duo E7600 @ 3.06 GHz (saved low/high temp differences):
Idle: 3 °C
Full load SETI@home: 7 °C

(CoreTemp tool)
ID: 1057278 · Report as offensive
-BeNt-
Avatar

Send message
Joined: 17 Oct 99
Posts: 1234
Credit: 10,116,112
RAC: 0
United States
Message 1057389 - Posted: 18 Dec 2010, 6:20:44 UTC

Yeah my e8400 had a stuck sensor that would not fall under 52c but when maxed out both core measure within 5-8 degrees of each other. My q9650's sensors are all working and can show up to a 10 difference idle but are within 8 degree's on load. Every computer I have ever built has been +/- 5-8 degrees if they had no stuck sensors. Hence the reason for delta and Thermal junction measurements.
Traveling through space at ~67,000mph!
ID: 1057389 · Report as offensive
Profile Francesco Forti
Avatar

Send message
Joined: 24 May 00
Posts: 334
Credit: 204,421,005
RAC: 15
Switzerland
Message 1057754 - Posted: 19 Dec 2010, 9:25:25 UTC

I was going to post something about temperature but before doing it I have tried to found if a thread was already opened on it.

I confirm that after the restart (week 40 and overall week 50) all my host are hotter. Previously all OK: temp <=10% than max. Now is > Max for 50% of time.
Increment is higher if CPUs are more than 2 (Max is on two 8 cores i7 that I have).

This on all my 8 hosts here under my controll (cuda or not).

I thing that it's caused by the particular type of data received.
I use (now and before) Win32 Lunatics' Unified Installer v0.37 in win 32 environment.

I can see, using Tthrotte Graph, that during the day temperature can vary about 5 C. depending not on the day/night cycle but on the task received.
This type of cyclic movement was also before but never over max.

FF
ID: 1057754 · Report as offensive

Message boards : Number crunching : Core Temperature Question


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.