Sign in with
Sign up | Sign in
Your question

Temps screwy with q6600

Tags:
Last response: in CPUs
Share
November 11, 2007 9:41:32 AM

Hi folks, quick question I have been able to find anything out about with google.

Just this morning put together a new system (my first intel one in 7 years) with a q6600 cpu, and asus p5n32e-sli plus mobo. Stock CPU cooler at the moment. Running XP pro SP2.

My problem is the cpu temps seem totally messed up, and I don't want to rely on them when I start overclocking the rig.

Using Asus PC probe it's giving me a mobo temp of 34 (probably about right) but cpu temp of 16 to 20C! This doesn't seem to change much whatever load I put on the sytem.

These readings are similar in the BIOS - CPU temp is sometimes 13C when I first boot up!

I've loaded the latest version of speedfan (4.33) and the temps it gives me are:

temp1 16C
temp2 34C
temp3 -55C (yep, minus 55!)
temp 1 -42C
temp 2 -38C
temp3 0C
CPU 0C
Internal temp 102C (!)
Remote temp 0C

I'mnot sure why there are 2 lots of numbers for temps 1 to 3, and i'd have though there'd be temps 1 to 4 if these are for each core separately, but i've not had a multicore CPU before.

Looks like there's a fault somewhere, is it fixable with some update (i've already got the latest bios), or does some hardware need swapping out? If so, is it likely to be CPU or mobo that's at fault?

Cheers folks
Jon

More about : temps screwy q6600

November 11, 2007 10:01:27 AM

First of all,you should try a MOTHERBOARD BIOS update to determine if that will fix the issue.As my first thought is a corrupt bios.But a BIOS update should fix the prob.If you already have the latest bios,then I would suggest trying a third party temp monitoring software.See what the readings from that are.If the problem still persists,try re-installing windows.If that fails,I recommend doing an RMA as it may very well be a faulty board.Goodluck.

Dahak

M2N32-SLI DELUXE WE
X2 5600+ STOCK (2.8GHZ)
2X1GIG DDR2 800 IN DC MODE
TOUGHPOWER 850WATT PSU
EVGA 7950 GX2 550/1400
SMILIDON RAIDMAX GAMING CASE
ACER 22IN WS LCD 1680X1050
250GIG HD/320GIG HD
G5 GAMING MOUSE
LOGITECH Z-5500 5.1 SURROUND SYSTEM
500WATS CONTINUOUS,1000 PEAK
WIN XP MCE SP2
Related resources
Can't find your answer ? Ask !
November 11, 2007 11:08:21 AM

As said, try this Core Temp. If it gives you screwy numbers then RMA the mobo, or possibly even the chip.
November 11, 2007 11:21:16 AM

Cheers guys, i've tried coretemp and it me much more sensible readings. 3 cores at 24 or 25, and 1 core at 30.

TAT didn't work for me - throws up the enumurator error message, which seems quite common.

Are there any good stressing programs that will stress all 4 cores? Prime95 doesn't seem to work all of them at all, neither does CPUburn - i'm after something that'll stick em all on 100%.
November 11, 2007 2:12:58 PM

jonboy99 said:
Cheers guys, i've tried coretemp and it me much more sensible readings. 3 cores at 24 or 25, and 1 core at 30.

TAT didn't work for me - throws up the enumurator error message, which seems quite common.

Are there any good stressing programs that will stress all 4 cores? Prime95 doesn't seem to work all of them at all, neither does CPUburn - i'm after something that'll stick em all on 100%.
You are using an old version of Prime95. You would need to run an instance per core. Try Prime95 25.5, it will stress all cores at one time.
November 11, 2007 3:56:09 PM

What kind of cooling are you using for your processor?

Those temperatures seem a bit low for stock air cooling unless you have way below 20°C ambient.
November 11, 2007 4:06:05 PM

Cheers zorg, prime95 cooking my quad right now :D 

Andrius, I was afraid you'd say that - it means i can't really rely on the temp monitoring at all then as i'm just using the stock cooler. However, i've just fitted a nice big zalman cnps7700 in there, with some arctic silver and temps are higher! Maybe too much AS, although i'm running it at minimum speed at the moment. Cores were sat just over 32 when I booted into windows, and are now 50, 50, 50, 54 after 5mins of prime and seem to have topped out.
A bit annoying that one core is always 3 or 4 degrees C higher than the others, would lapping improve this? Not sure I can really be bothered..
November 11, 2007 4:16:09 PM

Your temps are good, don't mess with it. I haven't ever seen all cores the same temp.
November 11, 2007 4:19:30 PM

This second set of temperatures is more like it (idle and load).

My Q6600 currently idles at 30°-32°-30°-24° @ 6x333MHz (8° difference).
It's below 20°C in my room so I think it's quite accurate (except for that 24°C). I use a Thermalright Ultra-120 Extreme cooler.
It's quite normal I think (since IHS & cooler base are not perfectly flat).
Lapping would help it.


My maximum load temepratures (after 1 to 15 hours of Prime95 @ 9x333) are 54°-54°-49°-47°C.




a b à CPUs
November 11, 2007 4:46:50 PM

jonboy99, if you want to know if your temps are correct, and learn how to uniformly test, accurately calibrate and properly monitor your temps, then check out the following Sticky: Core 2 Quad and Duo Temperature Guide - http://www.tomshardware.com/forum/221745-29-core-quad-t...

The Zalman 7700 is only on par with Intel's stock cooler, and is not up to the task of adequately cooling the Q6600, especially if it's B3 Stepping. Also, Atctic Silver 5 is not the best Thermal Interface Material.

From the Guide:

Scale 2: Quad
Q6600: Tcase Max 71c, G0 Stepping, Tjunction Max 100c, Vcore Max 1.372, TDP 95w, Delta 10c

-Tcase/Tjunction-
--70--/--80--80--80--80-- Hot
--65--/--75--75--75--75-- Warm
--60--/--70--70--70--70-- Safe
--25--/--35--35--35--35-- Cool

Scale 4: Quad

Q6600: Tcase Max 62c, B3 Stepping, Tjunction Max 100c, Vcore Max 1.372, TDP 105w, Delta 10c

-Tcase/Tjunction-
--60--/--70--70--70--70-- Hot
--55--/--65--65--65--65-- Warm
--50--/--60--60--60--60-- Safe
--25--/--35--35--35--35-- Cool

For information on CPU Coolers, please refer to the following link:

http://www.frostytech.com/articleview.cfm?articleid=204...

Masscool Shin-Etsu X23 Thermal Interface Material (TIM) can reduce CPU temps by ~ 4c. Although X23 is not well known, it is far superior to Arctic Silver 5, and is simply the best TIM for CPU cooling.

Shin-Etsu X23 is available at the following sites:

http://www.chillblast.com/product.php?productid=16932
http://www.ajigo-store.com/se7783d.html
http://www.crazypc.com/products/50118.html
http://www.watercoolingshop.com/catalog/product_info.ph... sCsid=78
http://www.specialtech.co.uk/spshop/customer/product.ph... t=0&page=

Comp :sol: 

November 11, 2007 7:16:09 PM

CompuTronix said:
The Zalman 7700 is only on par with Intel's stock cooler, and is not up to the task of adequately cooling the Q6600, especially if it's B3 Stepping.


jonboy99 said:
Cores were sat just over 32 when I booted into windows, and are now 50, 50, 50, 54 after 5mins of prime and seem to have topped out.
You are not satisfied with those temps? He is using the latest Core Temp 95.4, which shows the correct Tjunction without calibration. He is also using the latest Prime 95.4, which stresses all four cores. So his temps seem quite cool. Am I missing something here?
a b à CPUs
November 11, 2007 7:26:03 PM

Zorg, in the third sentence of jonboy99's opening post, he stated; "My problem is the cpu temps seem totally messed up, and I don't want to rely on them when I start overclocking the rig." Since he made it clear that he intends to overclock and will certainly be increasing Vcore, I'm sure that he would like to be aware of his limitations.

Comp :sol: 
November 11, 2007 7:42:59 PM

CompuTronix said:
Zorg, in the third sentence of jonboy99's opening post, he stated; "My problem is the cpu temps seem totally messed up, and I don't want to rely on them when I start overclocking the rig." Since he made it clear that he intends to overclock and will certainly be increasing Vcore, I'm sure that he would like to be aware of his limitations.

Comp :sol: 
That is possible, I OCed to 3G and actually lowered my Vcore from 1.3v to 1.275v. I guess it depends on how high he intends to OC.
a b à CPUs
November 11, 2007 8:11:59 PM

Yes.. and the characteristics of the individual Cores. With G0 Stepping and lapped A/C F7 Pro I have the following:

Ambient = 22c
Load = Prime95 Small FFT's 10 minutes

3.0Ghz
Vcore BIOS = 1.225
Vcore Load = 1.192
CPU (Tcase) = Idle 24c, Load 42c
Hottest Core (Tjunction) = Idle 33c, Load 52c

3.4Ghz:
Vcore BIOS = 1.375
Vcore Load = 1.344
CPU (Tcase) = Idle 26c, Load 52c
Hottest Core (Tjunction) = Idle 35c, Load 64c

3.6Ghz:
Vcore BIOS = 1.475
Vcore Load = 1.440
CPU (Tcase) = Idle 28c, Load 60c
Hottest Core (Tjunction) = Idle 37c, Load 72c

Comp :sol: 
November 11, 2007 9:26:00 PM

Excellent temps. I certainly agree, if he is looking at 3.6 or even 3.4 he will need to get a better cooler. He might even find his temps too warm for comfort at a mild 3G. Is that lap on the HS only, or both the HS and CPU? I should probably lap my Ultra120 Ex. the base is definitely not flat, of course neither is the IHS. I'm not really that concerned because I am just OCing to 3G, I want to keep this CPU for a while.
a b à CPUs
November 11, 2007 9:36:41 PM

I lapped the A/C F7 Pro because the heat sinc suface is quite course as manufactured compared to other coolers, and my particular unit was also quite distorted, and nowhere near flat. The IHS was just very slightly convex, but not nearly enough to justify voiding the warranty.
November 11, 2007 10:19:08 PM

I can't figure out why these HS manufacturers can't make the bases of these things flat and relatively smooth. They should add a couple of bucks it would be worth it. I did see a site, that was linked in a post, where you could have your Ultra120 lapped for a small fee, which is a great idea.
November 11, 2007 10:26:45 PM

SVC.com sells them handlapped to 2500 grit for an extra $40.
a b à CPUs
November 11, 2007 10:27:35 PM

I agree completely. Between travel time to purchase materials and labor time to remove, perform the lap, re-apply X23 and reinstall, I put in 5 hours. Of course, the result was well worth the time, however as you say, it would be great if all HSF manufacturers were required to meet flatness and smoothness specifications.

EDIT: cnumartyr, thank you very much for that information. I'll check it out.

Comp :sol: 
a b à CPUs
November 11, 2007 10:40:43 PM

Thanks...I'm already into it.
November 12, 2007 10:45:17 PM

Thanks for all replies folks, didn't realise people were still posting!

I have just found out about speedfan 4.33 having an offset error of 15degrees with cpus that are Tjunction Max 100c Stepping, because they expect Tjunction Max to be 85C and calculate the temp by subtracting a value from tjunction max.

I wonder if the bios (which is also reading the low temps like speedfan) is making this same error. Alas it is the most up to date BIOS for this board.

Googling 'Tjunction Max 100c Stepping speedfan' comes up with more stuff about this.
I've been running prime for about 40mins now and settled at coretemp temps of 57-60, and asus probe temps of 44-45. There always seems to be about 15 degrees difference between the two.

My CPU (according to cpu-z) is the GO stepping (this one: http://www.microdirect.co.uk/(21806)Intel-CPU-Core-2-Quad-Pro-Q6600-95W-Energy.aspx )

which is quite recent I think. Maybe a new BIOS needs to be released.
a b à CPUs
November 12, 2007 10:53:58 PM

jonboy99, I have already recommended that you refer to the Core 2 Quad and Duo Temperaturte Guide - http://www.tomshardware.com/forum/221745-29-core-quad-t... - which explains in detail the difference between CPU temperature and Core temperatures. There is a distinct difference between these temperatures, and the way in which they are acquired. Please check it out, including Section 9, which will allow you to "Calibrate" your temperatures so they are accurate and reliable.

Comp :sol: 
November 12, 2007 11:03:26 PM

Cheers comp, i'll check it out.

On a separate note, my mobo seems to suck for O/Cing - can't get it to 1200, even with multiplier at 7. So much for this being a good O/Cing board. :(  Southbridge a bit warm but all the heat pipes are well cooled by the zalman.
a b à CPUs
November 12, 2007 11:11:07 PM

Increasing Northbridge voltage will typically allow higher FSB frequencies.
November 15, 2007 3:35:33 AM

I guess you didn't try Core Temp that I posted earlier. It needs no calibration and will give you accurate core temperatures. I believe that the latest speed fan has been updated to read the correct Tjunction(max), hence needing no calibration either. I haven't checked this personally yet.
a b à CPUs
November 15, 2007 3:45:17 AM

Zorg, SpeedFan 4.32 and 4.33 both assume that Tjunction MAX is 85c, so it correctly interprets B0 Stepping only, and thus requires +15c Offset corrections for all Cores that are Tjunction MAX 100c. Beta 4.34 correctly interprets Tjunction MAX 85c and 100c, which applies to all Steppings, however, downloading Beta 4.34 requires users to jump through some registration hoops, which some may find too cumbersome with which to bother. I'm running Beta 4.34, and it works perfectly.

Comp :sol: 
November 15, 2007 4:02:59 AM

CompuTronix said:
Zorg, SpeedFan 4.32 and 4.33 both assume that Tjunction MAX is 85c, so it correctly interprets B0 Stepping only, and thus requires +15c Offset corrections for all Cores that are Tjunction MAX 100c. Beta 4.34 correctly interprets Tjunction MAX 85c and 100c, which applies to all Steppings, however, downloading Beta 4.34 requires users to jump through some registration hoops, which some may find too cumbersome with which to bother. I'm running Beta 4.34, and it works perfectly.

Comp :sol: 
I had pretty much steered clear of speedfan because I didn't trust what I was getting. Someone, I forgot who, told me that the new speedfan was accurate. Like I said, I haven't tried it yet and I'm in no rush. The core temp is small and gives me all I need which is the core temperatures. If I want more I use Rivatuner. I don't remember you commenting on Core Temp what do you think of it?

Edit: I just went back and looked at your guide again, I haven't read it in a long time. It appears your only problem with Core Temp is the confusion about the Tjunction being Tjunction max. I didn't have that confusion because I read the information on the website. I guess it pays to RTFM. Happy to see that is your only complaint.
a b à CPUs
November 15, 2007 4:18:19 AM

Zorg, I'm glad you asked. I think Core Temp is a sweet little program for experienced and informed users who understand temperatures and specifications, however, I have one HUGE problem with Core Temp, about which I've spoken to it's author, Arthur Liberman. He intends is to correct this problem in the next release, which is why I currently don't recommend it for monitoring. Below is a Note from my Temp Guide regarding Core Temp:

"...Section 8: Tools...

...Note 1: Core Temp is an excellent utility, however, it has a fundamental flaw in terminology, which creates confusion in the CPU temperature community, by obscuring the distinction between temperature and specification. Core Temp shows Tjunction 85c (or 100c), which is an incorrect term. The proper expression is Tjunction Max 85c (or 100c), which is the term defined by Intel as shown above in the Specifications section, and as represented below:

Tjunction = Core temperature
Tjunction Max = Shutdown

(A) Junction Temperature is a thermal measurement because it scales, thus the term TJ, or Tjunction, which is synonymous with Core temperature.

(B) Maximum Junction Temperature is a specification because it does not scale, thus the term TJ Max, or Tjunction Max, which is synonymous with Shutdown.

(C) ~ 5c below Tjunction Max Throttling is activated. If Tjunction Max is reached, Shutdown occurs, which is either 85c or 100c, and is determined by Stepping.

With this information in mind, it is recommended that Core Temp be used during Calibrations to cross-reference CPU-Z Stepping Revision and SpeedFan Core temperatures..."

Temperatures are confusing enough for the n00bs without adding this terminology problem into the equasion. I can't begin to count the number of posts I've seen from users who are confused by the "Tjunction 100c" field in the program, nor can I tell you how many times I've had to explain to confused users that it's just a specification and not a temperature, and that it doesn't change. Until Arthur Liberman fixes it, I think it's a pain in the a**.

Comp :sol: 
November 15, 2007 4:40:11 AM

I edited my post to indicate that I went back to your guide and looked. So other than the nomenclature, you like it? It should be pretty easy to add "max" in the GUI. I like it, it's small clean and to the point. I figure if the core temps are accurate and within tolerable levels then there is no need to get into Tcase etc. If you have a problem then you may need to delve a little further. What is your opinion?
a b à CPUs
November 15, 2007 5:04:57 AM

I agree. As I say, it's a sweet little program for advanced users and enthusiasts. I wouldn't recommend it for n00bs until they understand that Core temp scales with CPU temp at a known Delta. They can reach Tcase MAX and still have 20c to go before they reach Tjunction MAX, so Intel's spec's for Tcase MAX can be severely exceeded if they don't understand the thermal relationships. It's important to know the spec's for their variant, but many n00bs just don't bother to check it out. That's a good way to make silicon toast.

I really feel that the problem is with Intel's spec's. Tjunction MAX 100c just doesn't make sense. It's much too hot. The spec's for the B0 Stepping were more reasonable at 85c, which would work well for the new G0 and M0 Stepping processors. This would bring Tjunction to within throttle threshold at 80c when Tcase is nearing MAX at 70c. Sometimes I wonder what the engineers at Intel are thinking.
November 15, 2007 5:15:54 AM

Nice to know. I like to stay as far away from Tjunction (max) as possible. As you know I like a cool running CPU. I figure I can get several years out of this Q6600 before I am forced to upgrade.
a b à CPUs
November 15, 2007 5:26:30 AM

Gotta love these Core 2 processors. There hasn't been anything like this isnce the old Celeron 300A that would crank right up to 450.
November 15, 2007 11:20:23 AM

I missed that completely, I suspect I'm going to miss a lot more in the future. I probably won't upgrade until there is an IMC in every Intel CPU. Who knows, at that point maybe the AMD chip will be the one to have again.
November 15, 2007 11:30:34 AM

Zorg said:
I missed that completely, I suspect I'm going to miss a lot more in the future. I probably won't upgrade until there is an IMC in every Intel CPU. Who knows, at that point maybe the AMD chip will be the one to have again.


Late '08 / Early '09 will be Nehalem if Intel stays true.

Only thing AMD has on the roap map there is Fusion.

45nm seems the way to go considering the clocks on some of the Penryns.
November 15, 2007 2:24:39 PM

Zorg said:
I probably won't upgrade until there is an IMC in every Intel CPU. Who knows, at that point maybe the AMD chip will be the one to have again.
cnumartyr said:
Late '08 / Early '09 will be Nehalem if Intel stays true.

Only thing AMD has on the roap map there is Fusion.

45nm seems the way to go considering the clocks on some of the Penryns.
The point there is that I won't upgrade until there is an IMC in every Intel CPU. That won't happen for some time. According to what I have seen, Intel is only going to put the IMC in the high end chips to begin with and then later migrate the whole line to IMC. Sort of what they are doing/did with the single core CPUs. Put more bluntly, I'm going to ride my Q6600 until he!! freezes over. :lol: 
!