Sign in with
Sign up | Sign in
Your question

High core temps with e6750 and tjunction max at 100

Last response: in Overclocking
Share
November 18, 2007 11:48:49 PM

I composed a new system a while a ago including the following parts:

Lian Li PC-7 with side fan
Gigabyte 965p-ds3 rev. 3.3 (Bios F12)
Intel Core 2 Duo E6750 @ 3.2Ghz/400fsb (1.375v)
Kingston HX PC 6400 2x1GB DDRII 800 (4-4-4-12)
Asus EN8800GTS 320MB
Scythe Ninja Plus Rev. B (attached with arctic silver 5)
LC Power 550W Green Power
Samsung HD321KJ 320GB SATA-II
Samsung WriteMaster SH-S182M 18x

As you can see i set a slight oc of 20% - stresstested with prime for about 22 hours as well as tat for several hours. No crash, no errors no nada.
Temperatures i measured were also in range so i started to use it - which are about 4 moths now. I didn't encounter a single crash which wasn't caused by software incompatibilities(and no bsod).

Now recently Intel released the new tjunction max of 100 degrees for some of the c2d's - including mine. Several apps started to use that new tjunction value (Lavalys Everest, Speed Fan, Core Temp).

When first starting those apps i was quite surprised to see that my core temps were considerably higher than i previously assumed(around 15 degrees ofc).
I'm getting the following temps with the new tjunction max of 100 degrees (24 degrees room temp):
Idle: tcase temp: around 26 degrees, tjunction on each core: 42 - 46 degrees (which in core temp 0.954 beta is also around 55 - 60 degrees to tjunction max remaining)
Full Load under TAT: tcase temp: around 52 degrees, tjunction on each core: about 72 degrees (which in core temp 0.954 beta is also around 28 degrees to tjunction max remaining)
I get the same values with either Lavalys Everest or Speed Fan (newest version).

I have four questions considering those facts:

1. Are the tjunction (core) temps within a safe range on full loadand normal for idling? because i tend to think 72 degrees is quite hot on full load while around 43 idling...
2. Why do i have the big difference between tcase temp and tjunction (around 20 degrees compared to 10 degrees mentioned in the related thread)?
3. Is the jump of about 25 degrees (idling/full load with TAT) within specs?
4. If the first question is answered with yes: Is my heatsink properly attached - a scythe ninja seems so big it should cool the prcessor more effectively? Especially since i found lower idling and full load temps with the e6750 and core temp 0.954 beta...
4. If the first question is answered with no: Why is my system running stable for months now (gaming, hdtv etc)? Did i just reduce the lifespan of my processor dramatically?

I'd appreciate any info on this matter - i was digging through forums and irc chans for days now and did not find a conclusive answer to my problem and my head is starting to look like this: :pt1cable: 
thx in advance
November 20, 2007 1:33:57 PM

after i set the fan to constant full speed the temperature decreased by about 5 degrees in idling mode but only about 3 degrees on full load. However due to the lower idling temps the jump in core temp is now around 30 degrees from idle to full load (at max i get around 68-70 degrees).
Still too high in my opinion - i wonder if i just should reattach my heatsink. Then again 775 sockets are a bloody pain in the arse that's why i'm trying to use it as a last resort. Really would appreciate any help.
a b à CPUs
a b K Overclocking
November 20, 2007 5:54:13 PM

Caligu, I have taken the liberty of organizing your information into the following format, as shown in the Core 2 Quad and Duo Temperature Guide: http://www.tomshardware.com/forum/221745-29-core-quad-t...

"...Results

Tcase = Idle 26c / Load 46c
Tjunction = Idle 52c / Load 72c - (Hottest Core)

Variables

Ambient = 24c
Chipset = 965
C2Q / C2D = E6750
CPU Cooler = Scythe Ninja Plus Rev. B (AS5)
Frequency = 3.2 Ghz
Load = Prime95 - TAT
Motherboard = Gigabyte 965p-ds3 rev. 3.3 (BIOS F12)
Stepping = G0
Vcore = 1.375..."

From the Guide:

"...Section 6: Scale...

...Scale 1: Duo...

...E6x50: Tcase Max 72c, G0 Stepping, Tjunction Max 100c, Vcore Default 1.350, TDP 65w, Delta 10c...

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

From the Guide:

"...Section 8: Tools...

...Note 4: The software utility (TAT) Thermal Analysis Tool at maximum settings will simulate 100% Thermal Load, which would equate to Prime95 at 114% Workload. This provides the most extreme testing available for CPU and system cooling efficiency. Since TAT is coded to measure Notebook temps, it identifies a C2D as Pentium M. As Notebooks have no Integrated Heat Spreader, thermal scaling differs from desktops CPU`s, so TAT indicates ~ 2c lower, and depending on Variables, temps may be Offset by more than 15c. It is therefore recommended that TAT be used for extreme thermal testing only, and temps regarded as unreliable..."

From the Guide:

"...Section 9: Calibrations...

...Part 5: Test Setup - Load

Computer Case Covers = Installed
Computer Case Fans = Manual 100%
Computer Frequency = As Desired
CPU Fan = Manual 100%
CPU Internal Thermal Control = Enabled
Enhanced C1 Control (C1E) = Disabled
PECI (If Equipped) = Enabled
Speed Step (EIST) = Disabled
Vcore = Manual - As Required
Load Test = Prime95 - Small FFT's - 10 Minutes..."


The difference between TAT and Prime95 is about 5c, IF Prime95 is run as intended; Small FFT's for CPU Load testing, which is Prime95's heaviest processor Load, and is the same as Orthos. Remember that TAT's 100% Thermal Load (all one's through all registers), would be equivalent to Prime95 Small FFT's at 114% Work Load. TAT is the the most extreme CPU thermal testing available.

TAT exceeds Prime95 or Othos by 14%, and far exceeds any real-world video processing or gaming loads that would ever be applied to your processor, which is also true for Prime95 and Orthos. The most processor intensive apps and games running simultaneously would rarely reach a sustained 80% Work Load, so with this in mind, TAT represents an unrealistic CPU thermal test, except that it's useful to establish the extreme upper limits of processor and computer case cooling capacity.

Since your Tcase (CPU) appears to be too low relative to Tjunction (Core), and your Tcase to Tjunction Delta is inaccurate, I recommend that you follow the Calibrations procedure in Section 9, which specifies use of Prime95 - Small FFT's - 10 Minutes for the Load Test. When you've completed the Calibrations, please re-post your Results, as shown above.

Comp :sol: 
Related resources
November 20, 2007 6:22:28 PM

CompuTronix said:
Caligu, I have taken the liberty of organizing your information into the following format, as shown in the Core 2 Quad and Duo Temperature Guide: http://www.tomshardware.com/forum/221745-29-core-quad-t...

"...Results

Tcase = Idle 26c / Load 46c
Tjunction = Idle 52c / Load 72c - (Hottest Core)

Variables

Ambient = 24c
Chipset = 965
C2Q / C2D = E6750
CPU Cooler = Scythe Ninja Plus Rev. B (AS5)
Frequency = 3.2 Ghz
Load = Prime95 - TAT
Motherboard = Gigabyte 965p-ds3 rev. 3.3 (BIOS F12)
Stepping = G0
Vcore = 1.375..."

From the Guide:

"...Section 6: Scale...

...Scale 1: Duo...

...E6x50: Tcase Max 72c, G0 Stepping, Tjunction Max 100c, Vcore Default 1.350, TDP 65w, Delta 10c...

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

From the Guide:

"...Section 8: Tools...

...Note 4: The software utility (TAT) Thermal Analysis Tool at maximum settings will simulate 100% Thermal Load, which would equate to Prime95 at 114% Workload. This provides the most extreme testing available for CPU and system cooling efficiency. Since TAT is coded to measure Notebook temps, it identifies a C2D as Pentium M. As Notebooks have no Integrated Heat Spreader, thermal scaling differs from desktops CPU`s, so TAT indicates ~ 2c lower, and depending on Variables, temps may be Offset by more than 15c. It is therefore recommended that TAT be used for extreme thermal testing only, and temps regarded as unreliable..."

From the Guide:

"...Section 9: Calibrations...

...Part 5: Test Setup - Load

Computer Case Covers = Installed
Computer Case Fans = Manual 100%
Computer Frequency = As Desired
CPU Fan = Manual 100%
CPU Internal Thermal Control = Enabled
Enhanced C1 Control (C1E) = Disabled
PECI (If Equipped) = Enabled
Speed Step (EIST) = Disabled
Vcore = Manual - As Required
Load Test = Prime95 - Small FFT's - 10 Minutes..."


The difference between TAT and Prime95 is about 5c, IF Prime95 is run as intended; Small FFT's for CPU Load testing, which is Prime95's heaviest processor Load, and is the same as Orthos. Remember that TAT's 100% Thermal Load (all one's through all registers), would be equivalent to Prime95 Small FFT's at 114% Work Load. TAT is the the most extreme CPU thermal testing available.

TAT exceeds Prime95 or Othos by 14%, and far exceeds any real-world video processing or gaming loads that would ever be applied to your processor, which is also true for Prime95 and Orthos. The most processor intensive apps and games running simultaneously would rarely reach a sustained 80% Work Load, so with this in mind, TAT represents an unrealistic CPU thermal test, except that it's useful to establish the extreme upper limits of processor and computer case cooling capacity.

Since your Tcase (CPU) appears to be too low relative to Tjunction (Core), and your Tcase to Tjunction Delta is inaccurate, I recommend that you follow the Calibrations procedure in Section 9, which specifies use of Prime95 - Small FFT's - 10 Minutes for the Load Test. When you've completed the Calibrations, please re-post your Results, as shown above.

Comp :sol: 


Don't want to go offtopic but this is something I was wondering Comp:

Don't all the calibrations assume that the heatsink is properly mounted? Wouldn't tcase be significantly off if the heatsink was not making proper contact with the CPU causing innaccurate calibration results?
a b à CPUs
a b K Overclocking
November 20, 2007 7:31:22 PM

Daves255, thank you for your excellent questions. Yes, proper heatsinc mouning is required.

If the heatsinc is grossly mis-mounted, then post-calibration Idle Test results will typically indicate that Tcase to Tjunction Delta does not conform to Parameters. Load test results will typically indicate that Tjunction exceeds Hot Scale, and also indicates that Tcase to Tjunction Delta does not conform to Parameters, which suggests that a mounting problem exists.

Conversely, if a heatsinc is slightly mis-mounted, since the Idle Test Setup creates very low power consumption and thermal output conditions, then post-calibration Load Test results will may only indicate high Tcase to Tjunction Delta, but may not indicate excessive Tjunction, which suggests that a more subtle problem exists.

I will add a statement to the Guide on the next update to inform users that Load tests should be conducted to qualify temperatures and confirm proper heatsinc mounting prior to performing calibrations.

Thanks again for your thoughts,

Comp :sol: 
November 23, 2007 9:05:33 PM

First of all thanks a lot for your ellaborate reply - couldn't have hoped for more. Second, sorry it took so long but i had quite a busy week and was waiting for the weekend to continue testing.

Generally in this test setup i had a room temp of 22-23 degrees. Case closed. All fans on full speed.

The idling temps (with the cpu fan on full speed) are as followed:

22 degrees Tcase temp, tjunction1: 39 degrees, tjunction2: 40 degrees.


I ran Orthos Beta by Johnny Lee, which i assume has the same cpu load since it has the same small fft test as prime 95 but you only need to run it once. The temperature i got are as followed:

1min 30 secs: 49 degrees Tcase temp, tjunction1: 60 degrees, tjunction2: 60 degrees
6min 00 secs: 49 degrees Tcase temp, tjunction1: 62 degrees, tjunction2: 62 degrees
10min 0 secs: 50 degrees Tcase temp, tjunction1: 63 degrees, tjunction2: 63 degrees

All tjunction temps by «Core Temp 0.95.4» compared and verified with «Everest Ultimate Edition 4.20.1170», Tcase temps by «Everest Ultimate Edition 4.20.1170» compared and verified with «Speed Fan 4.34 beta 34».

The change from auto control to to full speed on the CPU fan has decreased idling temps by 5-7 degrees which might indicate a correct mount of the heatsink.

Interestingly the Tcase to Tjunction delta decreased from idling (17 - 18 degrees) to 13 degrees max at full load with Orthos Beta which might indicate inaccurate thermal diodes in the lower temperature range.

The high idle to full load delta remains high with a 28 degrees Tcase delta and a 25 degrees Tjunction delta which might indicate a slightly off mounted heat sink.

I added a pic of full load and idle:




It seems that I'm nowhere near a dangerous load now - still, as you mentioned in your second post, the doubt about a slightly mismounted heatsink remains...
a b à CPUs
a b K Overclocking
November 23, 2007 9:43:34 PM

Caligu, it always helps if you list your system Variables as shown above in my second post. Also, be sure that PECI (if equipped), APCI and ACIP support are enabled in BIOS, and the latest BIOS update should be downloaded and installed, since all of these items can affect Tcase linearity, and Tcase to Tjunction Delta.

Comp :sol: 
November 23, 2007 11:16:46 PM

I didn't change much tbh, also i found neither a PECI nor APCI or ACIP option in my gigabyte bios. It is also the latest bios released by gigabyte(F12):

Results

Tcase = Idle 22c / Load 49c
Tjunction = Idle 38c / Load 63c - (Hottest Core)

Variables

Ambient = 22c
Chipset = 965
C2Q / C2D = E6750
CPU Cooler = Scythe Ninja Plus Rev. B (AS5)
Frequency = 3.2 Ghz
Load = Orthos Beta
Motherboard = Gigabyte 965p-ds3 rev. 3.3 (BIOS F12)
Stepping = G0
Vcore = 1.375v"

I don't know if that helps...
November 23, 2007 11:18:40 PM

I forgot to mention - i posted the whole pics ofc - you need to right click "display picture" to see the whole image.
Sorry for double posting but the system won't let me edit my posts.
a b à CPUs
a b K Overclocking
November 23, 2007 11:39:41 PM

Caligu, use Section 9 in the Guide to Calibrate your temperatures.

Comp :sol: 
!