Sign in with
Sign up | Sign in
Your question

70c Core Temperatures - E3110

Tags:
Last response: in CPUs
Share
April 13, 2008 6:15:26 AM

My E3110 is showing as 70c for the core temperatures in CoreTemp, Hardware Monitor, Everest, etc.

Realtemp shows both sensors' movements at 0 or 1 (and 60c for core temps).

BIOS shows 30c idle temp for the cpu (DTS sensor).

Running prime95 does not increase the core temperatures at all (they ALWAYS stay between 72c-75c). It does, however, show an increase in the CPU (DTS sensor) temp from about 30c to 39c-40c.

I have a Zalman 9700 LED installed using AS 5.

Are the two core temperature sensors stuck? What are the chances of both of them being faulty? And does Intel warranty cover this? Thanks.
a c 122 à CPUs
April 13, 2008 8:09:39 AM

Quote:
Wasn't there something wrong with the 45nm core temp sensors


Actually it was with the E8400 only. And to tell you the truth what was happening with that, from what I have read, was the programs would end up reading the temp wrong since they were not yet calibrated to the new DTS sensor.

For the OP, did you try speedfan? Cuz at a temp of 70c it would more than likely start throttling or even shut down. I have a Q6600 @ 3GHz with the same HSF and mine will only hit 50c when I use Prime95. Might just be that Coretemp is too old. Do you have the latest version?

Might also be that that chip is a Xeon and not a C2Q so it might not be set to read it right.
April 13, 2008 8:18:47 AM

I have seen reports of the core temp sensors on the new 45nm chips not working.
Also, I believe the e3110 and the e8xxx series are essentially the same die.
Related resources
April 13, 2008 8:40:30 AM

I'm running with an E3110 as well. It only goes to 70c when I OC and run OCCT for some minutes.

Running OCCT now (no OC in idle) it shows temps at 43c and 39c while it shows 29c in AI suite/PC probe thingie. Same in SpeedFan around the 28-30c.
Just using Cooler Master Hyper TX2 on my CPU.

The thing is that even when it says 70c the cpu cooler is still cold to the touch, so I don't know.
a b à CPUs
April 13, 2008 9:42:55 AM

Darron said:
The thing is that even when it says 70c the cpu cooler is still cold to the touch, so I don't know.

Sounds to me like there is bad heatsink-CPU contact. That would explain the cold heatsink, because it isn't drawing the heat away from the CPU. You should reseat your cooler.

@demise14: Intel's warranty won't cover faulty sensors because at no time and in no place did they guarantee that the sensors will work. The only sensor that they care about is the IHS sensor or TCase (often called "CPU" in speedfan), which is always lower than the core temps. There's a reason they don't give out alot of details regarding maximum core temperatures.
April 13, 2008 9:46:59 AM

Darron said:
The thing is that even when it says 70c the cpu cooler is still cold to the touch, so I don't know.
I'm not saying this isn't a DTS problem, but you can't base the CPU temp on the HS temp, quite the opposite. The stock HS push pin mounting system is complete crap. If the pins are not securely locked in place the HS won't make proper contact with the CPU. The result is that the CPU can't transfer the heat to the HS. So the CPU gets hotter than he!! and the HS remains cool.

Of course this is completely unrelated to the wonky temps that some of the 45nm CPUs are reporting, even when using an aftermarket HS. It's just something to check if you have the stock HS.

Oops, randomizer beat me to it.
April 13, 2008 2:48:05 PM

So I guess my main question is: Is it possible that these programs can be showing the correct core temperatures? I've reinstalled the HSF numerous times using different pastes, reverted back to the stock HSF (which gives 45c idle temps on the DTS sensor), and my case has very good cooling.

Any other way to test the core temperatures? I installed Speedfan and played around with it for a minute. The only temperature it showed was the DTS sensor (which seems to fine at the moment).
April 13, 2008 4:25:30 PM

Check to see if there's an update for your bios. They may have a beta to try out.

I saw another thread on this and saw Gigabyte had beta bios for wolfdales, and the latest had yorkies. So you never know, could be a bios thing. :D 
April 13, 2008 5:07:57 PM

^ I've already updated to F3 bios.
April 13, 2008 5:27:59 PM

Take the heatsink off, clean both the surface of the CPU and the Surface of the Heatsink, reapply a bit of thermalpaste, just a pea size if not smaller, then give it a whirl that way.

Make sure the Heatsink is really on there.. nice and tight

If you already did this, then sorry lol
April 13, 2008 6:00:13 PM

If it was me, I would put a quality aftermarket HS, with a back plate, on there and if the temps didn't come down then at least I wouldn't have to worry about the stock HS pins. IMO don't get an aftermarket HS with the push pins.

Additionally, CompuTronix updated his guide to get a better handle on the CPU temperature of offset, read it.

Core 2 Quad and Duo Temperature Guide
April 13, 2008 7:29:58 PM

It only goes that high when I OC it though, so no biggie for my normal use.
I did clean both cpu and cooler using isopropylalcohol 99% (I'm a Dane, so might spell differently in US/UK) and I'm 100 % the cooler sits correctly.
I used Arctic Silver 3 on it.
I'll take it apart and have a look tomorrow and post back :) 
a c 122 à CPUs
April 13, 2008 8:20:21 PM

Zorg said:
If it was me, I would put a quality aftermarket HS, with a back plate, on there and if the temps didn't come down then at least I wouldn't have to worry about the stock HS pins. IMO don't get an aftermarket HS with the push pins.

Additionally, CompuTronix updated his guide to get a better handle on the CPU temperature of offset, read it.

Core 2 Quad and Duo Temperature Guide


Um he is using a Zalman CPNS9700 which uses a bolt through backplate and a screw system to mount the HSF on the backplate thingy. I know cuz I have one. And trust me its hard not to have that thing on properly.

I still think the should get the most recent version of CoreTemp and try SpeedFan.
April 13, 2008 8:47:11 PM

Oops, I see it now.

Yeah, the Zalman with the backplate is probably properly mounted. I wouldn't take it back off to check, just leave it alone. The problem that the you are having is probably the offset which is addressed in the temp guide.
a b à CPUs
April 13, 2008 8:47:38 PM

Zorg said:
If it was me, I would put a quality aftermarket HS, with a back plate, on there and if the temps didn't come down then at least I wouldn't have to worry about the stock HS pins. IMO don't get an aftermarket HS with the push pins.

Additionally, CompuTronix updated his guide to get a better handle on the CPU temperature of offset, read it.

Core 2 Quad and Duo Temperature Guide

Agreed. Well said on the guide.
a b à CPUs
April 13, 2008 8:49:29 PM

Quote:
Wasn't there something wrong with the 45nm core temp sensors

Agreed. Btw, E8400 = E3110. Both OC to the same level and are based on the same core. Only difference is that E3110 might be of a "better bin" since it is a Xeon chip.
April 14, 2008 3:31:20 AM

Quote:
^Yeah kinda like those 75W opterons that float around newegg.
Hey if its a better bin then wouldn't it have better sensors and a higher Q.C?


they are the same chip just branded as a Xeon. Unless there's a clear explanation proving me wrong it's the same chip, same bin, and the same core. Only supports uni-processor. There is nothing special about it.
April 14, 2008 12:52:45 PM

^^goodo
a b à CPUs
April 14, 2008 9:55:54 PM

Quote:
A heathly diet creates humor


Im confused... mind explaining? :pt1cable:  :pt1cable:  :pt1cable: 
a b à CPUs
April 15, 2008 11:01:03 AM

^Ah... :lol: 
April 16, 2008 4:53:18 PM

My email convo with the retailer I purchased from: (Note: first email is on the bottom, last sent email is on the top)

Quote:
I could completely agree with you if wasn't for the fact that the OTHER units have core temp sensors that are working correctly. The thing is, as far as I know, my processor is the only one that is giving 70c temp readings from the core sensors. I've searched and searched, and some people have readings that are in the 50's, 60's, etc. under LOAD, but 70c is my idle temp.

If every chip was shipped with broken sensors, then, like I said, I'd completely understand you. But there are E3110's and E8400's with sensors that are working just fine. Tell me, how is it fair that one person pays $220 and gets a 'complete' product, while another person pays the same amount but receives a product with broken parts.

Core temperatures can run hot while the chip sensor shows relatively low degrees (or else, why would they exist in the first place?). Again, thanks for the quick reply and good argument. I'm not trying to sound petty and I'm not trying to overreact. But I know I'd regret it if I didn't try to get a return/exchange for an incomplete product. It's just hard building a new machine and coping with that "it's not complete/it has problems" feeling rather than the feeling you get when everything is working fine, you know you good deals, and you're satisfied with all of the components. If you're telling me that a return is not possible, then fine, there's not much I can about it other than sticking with NewEgg next time, I suppose. Thanks again.



--------------------------------------------------------------------------------
Date: Mon, 14 Apr 2008 08:41:38 -0400
From: ***********
To: ***********
Subject: Re: Order Question

Hi demise14-

If it overheats and dies, obviously then you'd be able to get an exchange. The thing is, the throttling and emergency shutdown on the chip are controlled by the chip temp sensor that the BIOS reads - not the core temps. So your chip won't overheat and kill itself. This is part of why Intel isn't willing to do exchanges on them - their position is that the chip works fine, won't overheat, and the important temp sensor works fine. Intel isn't even releasing official software to read the core temps - it's not something they actually support. Intel previously released a temperature monitoring software for the 65nm chips that did reliably read the core temps, but they have not extended that support to the 45nm chips, and to my knowledge, do not intend to.

To address your car/AC analogy, there is one point missing. The thermostats that are actually important, supported by the companies, and actually control the systems work fine. The issue is when you install a 3rd party monitor that ties into an unsupported/unimportant (in their eyes) sensor. In either case, they'd tell you the same thing - the system they actually support works just fine. While knowing the core temps was nice, helpful information, it's simply not supported anymore, and Intel doesn't care if it doesn't work.

As I have told other people with similar issues, it boils down to this: The chip is functional in Intel's eyes. It runs at the specified speed, within the thermal specifications they defined. So long as the Tcase sensor shows a temp less than 74.1 degrees, it's working fine. Since they won't do a return/exchange, we're really stuck. As much as I'd love to override that, we just don't have a choice. Clearly, if you find the chip actually does fail, we'd most certainly press for a replacement without issue.

While I wish this weren't an issue, and while I may not necessarily agree with Intel's position on this, please understand it's just that - Intel's policy/position, and we've got no choice but to follow it. I am also still talking with our Intel rep to see if we have any other options, and if we do get word from them, I'll most certainly let you know. For the time being, however, the processor is considered by Intel to be fully functional, and ineligible for an exchange.

-******

demise14 wrote:
Understandable. However, if my CPU overheats, who is responsible then?

You can't safely use your car without having the engine temperature gauge. And I can't safely use my processor without knowing the core temperatures.

I see you've assumed that this is only an issue because you think I intend on overclocking the unit. Regardless, I like my PC to be kept cool and it's very unsatisfying when I purchase a $50 aftermarket HSF, but can't even see how that money made a difference in the temperatures. If this isn't considered important, then what exactly is covered by the warranty?

Thanks for the reply. I can see how an exchange on such a rare item (at the moment) is difficult, but why is a refund not acceptable?

Honestly. If you purchased a home air conditioning system and the temperature gauges were faulty, how would you feel if the company you purchased it from told you that this isn't considered a problem and a refund is out of the question?



--------------------------------------------------------------------------------
Date: Sun, 13 Apr 2008 18:57:09 -0400
From: ***********
To: ***********
Subject: Re: Order Question

Hi demise14-

The temp in the BIOS is the one Intel considers important. The core temps are not something they really "support", and they will tell you to ignore them. This is a fairly widely reported "problem" which Intel says is not really a problem at all. Unfortunately, they are not accepting any returns/exchanges on these processors, which means that we can't either. In Intel's eyes, the only important temperature is the BIOS reading, which is the one used to throttle/shut down the machine. Of course, I naturally understand that not being able to easily read core temps makes overclocking very difficult - but unfortunately overclocking isn't something that's warrantied anyway. As much as I'd love to be able to take exchanges on these chips, since Intel won't, we can't either. If you have any other questions though, please don't hesitate to ask!

-******


On Sat, Apr 12, 2008 at 11:19 PM, demise14 wrote:

Hi again.

The E3110 I purchased has faulty DTS (core) sensors.
The core sensors both show as 72c degrees idle (CoreTemp).
Realtemp and CPUZ Hardware Monitor are showing the core temperatures as 62c, idle.
Bios (which reads off of the Tcase sensor) is showing 30c, idle (this is the sensor between the two cores - seems to be working fine).

Realtemp's sensor test shows both core sensors as stuck. The HSF is correctly installed (aftermarket Zalman 9700 cooler) with AS 5 thermal paste.

How would I go about getting a replacement or refund?
a b à CPUs
April 17, 2008 12:46:22 AM

That retailer knows what he's talking about, you don't. I wish we had retailers like that here. Don't you get it? Your CPU is not faulty because it operates exactly the way Intel advertises and intends it to operate.
April 17, 2008 12:53:12 AM

Sorry to say, your retailer's right on this one. They have said that if the chip overheats and fails, they'll replace it; I'm not sure what more they could be reasonably expected to offer in this case. When you buy a chip, you can't really complain to the vendor if something that isn't officially part of the chip doesn't happen to exist on your chip.
April 17, 2008 9:21:16 PM

I know he's right to some degree. But it was still worth a shot on trying to get a replacement. Complaining can be very effective sometimes.

But I still stand behind my point here: "Tell me, how is it fair that one person pays $220 and gets a 'complete' product, while another person pays the same amount but receives a product with broken parts."

But oh well, I can live with guestimating my core temps. My DTS sensor shows 22c idle now that the thermal paste has had time to cure. So I'm happy.
a b à CPUs
April 17, 2008 10:18:45 PM

demise14 said:
I know he's right to some degree. But it was still worth a shot on trying to get a replacement. Complaining can be very effective sometimes.

But I still stand behind my point here: "Tell me, how is it fair that one person pays $220 and gets a 'complete' product, while another person pays the same amount but receives a product with broken parts."

But oh well, I can live with guestimating my core temps. My DTS sensor shows 22c idle now that the thermal paste has had time to cure. So I'm happy.

Both of you are getting the same product. Depending on the motherboards manufacture they would have a BIOS update to fix the DTS problem. There is a workaround for your problem. You could:
1. Update BIOS and see if it fixes it.
2. You can calibrate the CPU temps according to Core 2 Quad and Duo Temperature Guide . See under Section 9.
a b à CPUs
April 17, 2008 10:19:53 PM

randomizer said:
That retailer knows what he's talking about, you don't. I wish we had retailers like that here. Don't you get it? Your CPU is not faulty because it operates exactly the way Intel advertises and intends it to operate.

Agreed.
April 18, 2008 3:43:08 AM

demise14 said:
...
But I still stand behind my point here: "Tell me, how is it fair that one person pays $220 and gets a 'complete' product, while another person pays the same amount but receives a product with broken parts."...

The "broken parts" aren't officially part of the product you were buying, so it doesn't matter if yours has them or not -- you each got what you paid for.


a b à CPUs
April 18, 2008 9:03:38 AM

Why not next time complain that the bubble wrap on your shipment that you paid for was partly popped when somebody else's wasn't? Same deal. It's not a part that is required for the intended functionality of the product to be fully utilised.

Other option: Don't buy CPUs known to be "incomplete" :heink: 
April 18, 2008 9:06:22 AM

Seems I misunderstood the original question, as my temps aren't "fixed".
I only get that high when I OC and stuff.
April 18, 2008 11:32:54 PM

Bios calibration/updates are completely irrelevant as core temps are not dependent on bios calibration, calibrated at factory, and can be read directly from 7 bit register on cpu. Cpu temps are diodes that must be calibrated and thus require bios support/calibration.

Intel states in their white papers, "Each individual processor is calibrated so that TCC activation occurs at a DTS value of 0. The temperature reported by the DTS is the relative offset in PECI counts below the onset of the TCC activation and hence is negative. Changes in PECI counts are roughly linear in relation to temperature changes in degrees Celsius. For example, a change in PECI count by '1' represents a change in temperature of approximately 1°C. However, this linearity cannot be guaranteed as the offset below TCC activation exceeds 20-30 PECI counts.

Demise14, you have the worst stuck/nonlinear sensors I have seen. There is no fix.

Intel in past, if dealt with them directly, have replaced cpu's with faulty sensors. However, in past, faulty sensors were the exception not the rule, hence the shift in policy.

I agree with intel to a point, if your core temp is reading 40 or 50, fine, but yours is suggesting ~30 from tjmax at idle, that is not only the worst I have seen, but it is ridiculous.

You would have two avenues of returning the chip...and go through intel, not the innocent, powerless middle man.
1) If your cpu does not throttle at DTS=0, or if your sensors can not reach DTS=0, (but you would have to take risk of running your chip to DTS=0 and monitor with Fluke to make sure cpu did not fry if sensor does not work.)

2) from intel "TCONTROL can be described as a trigger point for fan speed control implementation. The processor TCONTROL value provided by the Digital Thermal Sensor is relative and no longer absolute. The TCONTROL value is now defined as a relative value to the TCC activation set point (i.e. PECI Count = 0)" In other words, if your delta to tjmax is reporting 35 to TCC at idle, your fans may be driven at higher speed all the time, making a quiet solution impossible, this deviates from design function.

But I would contact intel directly, and see what they have to say.

Intel needs to draw the line somewhere granted, and I would accept the nonlinear issue or core reading of 50 idle or one stuck sensor. But yours, especially if it interferes with tcontrol, a cpu that reads ~30 from throttling at idle, intel has completely lost their mind if they think that is acceptable.

a b à CPUs
April 18, 2008 11:48:48 PM

rge said:
Bios calibration/updates are completely irrelevant as core temps are not dependent on bios calibration, calibrated at factory, and can be read directly from 7 bit register on cpu. Cpu temps are diodes that must be calibrated and thus require bios support/calibration.

Intel states in their white papers, "Each individual processor is calibrated so that TCC activation occurs at a DTS value of 0. The temperature reported by the DTS is the relative offset in PECI counts below the onset of the TCC activation and hence is negative. Changes in PECI counts are roughly linear in relation to temperature changes in degrees Celsius. For example, a change in PECI count by '1' represents a change in temperature of approximately 1°C. However, this linearity cannot be guaranteed as the offset below TCC activation exceeds 20-30 PECI counts.

Demise14, you have the worst stuck/nonlinear sensors I have seen. There is no fix.

Intel in past, if dealt with them directly, have replaced cpu's with faulty sensors. However, in past, faulty sensors were the exception not the rule, hence the shift in policy.

I agree with intel to a point, if your core temp is reading 40 or 50, fine, but yours is suggesting ~30 from tjmax at idle, that is not only the worst I have seen, but it is ridiculous.

You would have two avenues of returning the chip...and go through intel, not the innocent, powerless middle man.
1) If your cpu does not throttle at DTS=0, or if your sensors can not reach DTS=0, (but you would have to take risk of running your chip to DTS=0 and monitor with Fluke to make sure cpu did not fry if sensor does not work.)

2) from intel "TCONTROL can be described as a trigger point for fan speed control implementation. The processor TCONTROL value provided by the Digital Thermal Sensor is relative and no longer absolute. The TCONTROL value is now defined as a relative value to the TCC activation set point (i.e. PECI Count = 0)" In other words, if your delta to tjmax is reporting 35 to TCC at idle, your fans may be driven at higher speed all the time, making a quiet solution impossible, this deviates from design function.

But I would contact intel directly, and see what they have to say.

Intel needs to draw the line somewhere granted, and I would accept the nonlinear issue or core reading of 50 idle or one stuck sensor. But yours, especially if it interferes with tcontrol, a cpu that reads ~30 from throttling at idle, intel has completely lost their mind if they think that is acceptable.

I know CoreTemp doesn't rely on BIOS for the DTS reading. The problem is that the DTS that is reported by the E8x00 CPUs are incorrect.
a b à CPUs
April 18, 2008 11:52:43 PM

@OP: PM CompuTronix and see what he says. He knows best whenit comes to Temps.
April 19, 2008 2:29:14 AM

Shadow703793 said:
I know CoreTemp doesn't rely on BIOS for the DTS reading. The problem is that the DTS that is reported by the E8x00 CPUs are incorrect.


"The DTS that is reported by the E8x00 cpus are incorrect"???. Are you trying to say that coretemp is reporting DTS output incorrectly?
If so, not true. If you are saying DTS output is inaccurate on some intel chips, then I would agree.

Regardless, as most know the Digital Thermal Sensor's logic scans several transistors functioning as diodes on die and reports the maximum temp as the core temp for each die, and this temp is stored in register in 7 bit digital output, that is easily and accurately read by most software. Coretemp, realtemp, everest, hardware monitor, all are accurately reading this delta to throttling temp number, and all are reporting it exactly the same. (They only give absolute temps varying by 5-10C based on different tjmax's). http://download.intel.com/technology/eep/fall_microproc... see slide 22 DTS sensor function. Intel has confirmed in their forum, now locked, that it is software readable and how.

Saying, DTS (raw data, not absolute temp) is being reported incorrectly by software is just like saying it needs to be calibrated by bios, both are false. Read this thread. http://www.xtremesystems.org/forums/showthread.php?t=17...

And if you do not believe intel documentation, unclewebb and I and others have used flukes to measure temps and posted pictures on xtreme showing linear scaling of this delta to tjmax (reported accurately by all of the programs) that occurs on those with properly functioning sensors (such as mine) providing the temp is around 60C and up, and temp scales linearly at that point.

OP problem cpu has absolutely nothing to do with "not being read properly" or "3rd party programs cant read it". 3rd party programs read delta to tjmax exactly correct. (Though converting to absolutes is guessing.) But the OP is 35C from tjmax at idle, that is not software reading something incorrectly, that is his sensors being unacceptably inaccurate at a point that intel states in their own white papers that they should be accurate.

National Semiconductor has illustrated some of problems of 45nm DTS accuracy problems, using their competitors that apparently report significant temp inaccuracies http://www.national.com/appinfo/temp.../trutherm.html
While obviously biased, it does offer the point of more problematic 45nm temp reporting.

Edit:
And if anyone wants to read the raw data output of DTS and check against temp probe or fluke measurement themselves, so you know yourself instead of relying on others opinions. THOUGH BE CAREFUL IF YOU DO THIS, AND DO SO AT OWN RISK.

1) download crystal cpuid here... http://crystalmark.info/download/index-e.html#CrystalCP...
2) Run Function>>MSR editor. Then enter 0x19c in the MSR number (and read and heed the warning). Click on RDMSR (read model specific register) (NOTE DO NOT HIT WRMSR which is write model specific register unless you want to discover the hard way whether that register is write protected ro not)...you will get number in EAX of something like 0x88390000. Read two entries following the 0x88, in this case 39.
3) then convert this hexadecimal output to decimal, using either the windows calculator in scientific mode (hit view change scientific) or go here and use tables http://www.ascii.cl/conversion.htm, 39 converted to decimal is 57. So I am 57 units (not degrees C) from TCC throttling. These DTS ouput units are "guaranteed to be linear and 1 unit equal 1 C when temp is 20-30 unit from TCC, they are clearly not linear at lower temps, and 45nm are the most nonlinear cpus yet at lower range.

Full documentation of reading MSR from intel....http://download.intel.com/design/processor/manuals/2536...
quote
"13.5.5 On Die Digital Thermal Sensors
On die digital thermal sensor can be read using an MSR (no I/O interface). In Intel Core Duo processors, each core has a unique digital sensor whose temperature is accessible using an MSR. The digital thermal sensor is the preferred method for reading the die temperature because (a) it is located closer to the hottest portions of the die, (b) it enables software to accurately track the die temperature and the potential activation of thermal throttling."
a b à CPUs
April 19, 2008 4:05:55 PM

rge said:
"The DTS that is reported by the E8x00 cpus are incorrect"???. Are you trying to say that coretemp is reporting DTS output incorrectly?
If so, not true. If you are saying DTS output is inaccurate on some intel chips, then I would agree.

Yes. That is what I meant ;) .
!