Real Temp - New temp program for Intel Core processors

[lunadesign]

I'll have to check up on this. I thought it was reading a sensor that was located in the CPU socket but don't quote me on that.

Looks like the sensors on core0 and core1 are both working fine. My best guess is that your actual core temperature is between 28C and 30C during your test. If you want to keep things simple you could assume 28C so you wouldn't need to do any calibration on core1 and then you could use a -1.0 calibration on core0 which should bring it down to 28C also. If you choose to go up a degree or two, that's fine. Just use calibration factors so core0 and core1 are balanced during this test. I like the simple approach where you only have to use one factor. When you return to your normal MHz, you might need to do a slight adjustment. Core0 and core1 should be pretty much equal at idle even when overclocked and overvolted.

I like using Prime95 v25.6 running small FFTs. Are you using a multi-core version of Prime or running 4 individual instances of Prime? I'm not sure if this would make any difference to your results. At full load I generally find that core0 and core1 are equal while core2 and core3 are equal but the two sets of cores might be off by a few degrees. One user I helped had load temps similar to yours where the two outside cores ran hotter than the two inside cores and I think when he re-did his thermal paste this changed. Anything like installation procedure or an IHS / heatsink that is not perfectly flat can cause slight differences at full load as well as less than perfect sensors. Probably not worth worrying too much about though.

The other test I use is to use Prime95 and to go into the task manager and set the affinity to limit it to run on 1, 2 or 3 of the 4 cores at a time and to switch the load around while Prime is running. If you get some interesting numbers then post them.

I don't believe the Xeon sensors are any different. The 45nm sensors just aren't always capable of giving us the information that we'd like to be able to get out of them. It's sort of random luck and I think you're doing better than most users have. At least you took the time and did your homework so you know exactly what your sensors are capable of and the temperature range where they give you accurate data.

Thanks for the info! I ended up having to use a -1.8 calibration on Core0 to get it roughly matched up with Core1.

To answer your question, I generally run 4 instances of Prime95 and use the Affinity (-A) option to tie each instance to its own core. Is there a multi-core version of Prime95 out there? That would make it much easier...

I think I'm going to take your advice and not return the CPU...too much hassle especially if I'm likely to get another one like this. I'll be building a few more systems with almost identical specs so it will be interesting to see how the next ones match up with this one.

[CompuTronix]

Originally Posted by Lunadesign
Regarding the CPU temp that's reading 20C (Tcase), I thought this sensor was located on the case of the processor but read by the motherboard?

I'll have to check up on this. I thought it was reading a sensor that was located in the CPU socket but don't quote me on that.

Unclewebb & lunadesign,

This is a very common misunderstanding, and is precisely why users hesitate to trust Intel's Tcase sensor, which is the processor-specific thermal specification shown in Intel's Processor Spec Finder - http://processorfinder.intel.com/det...px?sSpec=SLAWQ

The era of the thermocouple-in-the-center-of-the-CPU-socket has long since passed.

The following Intel document - http://arxiv.org/ftp/arxiv/papers/0709/0709.1861.pdf - clearly shows on Page 2, Figure 1, upper right hand corner, the "Analog Sensor" which is embedded within the substrate layers of the processor package. Excerpts from my Core 2 Quad and Duo Temperature Guide - http://www.tomshardware.com/forum/22...perature-guide - explain how this Tcase sensor has worked for the past several years, which actually applies to some previous generations of Intel, as well as AMD processors:


"Section 1: Introduction

Core 2 Quad and Duo processors have 2 different types of temperature sensors; a CPU Case (not computer case) Thermal Diode located within the CPU die between the Cores, and Digital Thermal Sensors located within each Core...

Section 3: Interpretation

The first part of the spec refers to a measuring point on the Integrated Heat Spreader (IHS). Since a thermocouple is embedded in the IHS for lab tests only, IHS temperature is replicated using a CPU Case Thermal Diode integrated between the Cores. Maximum Case Temperature is determined by Spec#. The CPU Case Thermal Diode is how Tcase is measured, and is the CPU temperature displayed in BIOS and the software utility SpeedFan...

Section 5: Findings

(A) Tcase is acquired on the CPU Die from the CPU Case Thermal Diode as an analog level, which is converted to a digital value by the super I/O chip on the motherboard. The digital value is BIOS Calibrated and displayed by temperature software. BIOS Calibrations affect the accuracy of Tcase, or CPU temperature.

(B) Tjunction is acquired within the Cores from Thermal Diodes as analog levels, which are converted to digital values by the Digital Thermal Sensors (DTS) within each Core. The digital values are Factory Calibrated and displayed by temperature software. Factory Calibrations affect the accuracy of Tjunction, or Core temperatures.

(C) Tcase and Tjunction are both acquired from Thermal Diodes. Tcase and Tjunction analog to digital (A to D) conversions are executed by separate devices in different locations. BIOS Calibrations from motherboard manufacturers, Factory Calibrations from Intel, and popular temperature utilities are frequently inaccurate.

(D) Intel shows Maximum Case Temperature (Tcase Max) in the Processor Spec Finder, which is the only temperature that Intel supports on Core 2 desktop processors. Ambient to Tcase Delta has known Offsets which vary with power dissipation and cooler efficiency, and can be Calibrated at Idle using a standardized Test Setup..."


Since the first release of Core Temp, the overclocking community has become so brainwashed on core temperatures, that they now overlook CPU temperature as a reliable thermal measurement, let alone as a secondary reference. I have yet to see a Tcase sensor "stick", and as Intel has designed this sensor specifically for temperature measurements and depends on it's accuracy, my experience and observations is that the Tcase sensor scales with linear characteristics.

The only problem with the Tcase sensor is that BIOS programmers are confined to "canned" values, and sometimes incorrectly code just one of the many Socket 775 variants into BIOS. Regardless of whether BIOS is correctly offset for a particular processor, Tcase can still be calibrated in SpeedFan to an accuracy of within a degree or two. By using a similar standardized test setup for low Vcore and frequency, with covers removed and fans at 100% RPM, if ambient is accurately measured, then idle power dissipation and CPU cooler thermal efficiency are easily calculated to provide accurate CPU temperature.

I Hope this helps to clear things up.

Comp

[rge]

Providing you have a modern motherboad, and most do, cpu temp is indeed read from a diode between the cores. Computronix, I agree with above, except the semantics issue of calling cpu diode a Tcase sensor. Intel will tell you to use the cpu diode to monitor Tcase specs, but it is not actually Tcase, and when you call them on it, they punt. There can be, mathematically, as much as ~15C difference between cpu diode and a sensor placed on IHS under certain testing full load TDP parameters, using intel formulas. I went round and round via email with intel via cpu temp vs Tcase (actual Tcase is IHS where no sensor exists). I already posted most of emails on anandtech, it is here. http://forums.anandtech.com/messagev...&enterthread=y

gradient goes from 1) core (~120W/m*k thermal conductance) to 2)between core (still in die material and still ~120W/m*k) and thus max gradient between core and cpu temp is ~5C on load.

gradient continues from between cores to 3) die attach solder ? ~10W/m*k (from companies that adverstise) 4) IHS over ~120w/m*k, hence most of the gradient occurs at the die attach adhesive.

So although only 5C max gradient occurs from core to cpu across 120w/m*k conductance material, a possible ?23C gradient exists under certain load, max TDP, intel cooling conditions from core to IHS center across lower conductance solder attach ~10w/m*k (if measured with sensor on IHS). (solder ball in pic below is not same as solder attach, for those looking at that)

see anandtech link for reasoning.

pic of common intel conductances.

[CompuTronix]

Computronix, I agree with above, except the semantics issue of calling cpu diode a Tcase sensor. Intel will tell you to use the cpu diode to monitor Tcase specs, but it is not actually Tcase, and when you call them on it, they punt. There can be, mathematically, as much as ~15C difference between cpu diode and a sensor placed on IHS under certain testing full load TDP parameters, using intel formulas. I went round and round via email with intel via cpu temp vs Tcase (actual Tcase is IHS where no sensor exists). I already posted most of emails on anandtech, it is here. http://forums.anandtech.com/messagev...&enterthread=y

rge, I quite clearly understand the definitions, and the major and minor variables, as well as the fine points of where the temperature is actually measured in the lab, as compared to how the end user's "CPU temperature" is "replicated".

Also, I've read that Anandtech thread, and as in so many other Forum threads, I'm continually amazed by how vague, evasive and ambiguous Intel's responses typically are. It's like asking the CIA to provide us with the First Lady's recipe for chocolate chip cookies obtained through the Secret Service.

Like unclewebb, I wonder sometimes if the guys in the blue clean room bunny suits with space helmets are going to show up in the wee hours of the night and abduct us for trying to expose the truth about Core 2 temps!

Comp

[rge]

I figured you knew the difference, which is why I called it a semantic issue, my explanation was more for others trying to follow my logic.

As we both know, intel is using one term to describe two different points, in essence oversimplifying error either from laziness or intentional vagueness, or, giving intel the benefit of the doubt, if you are in specs via cpu temp by definition you are within Tcase/IHS temp spec (since that would be even lower). If you rewrite your guide, could you make me happy and differentiate the two, since intel is too lazy or secretive to do so?

[CompuTronix]

If you rewrite your guide, could you make me happy and differentiate the two, since intel is too lazy or secretive to do so?

I'd be happy to do that, if it were only advanced overclocking enthusiasts reading the Guide, but others have mentioned that it's too complicated.

I'm inclined to keep it simple in this instance.

Comp

[unclewebb]

Thanks for the additional info. I thought there was a separate sensor but I also read that Intel had removed this additional temperature sensor on some of their newer CPUs. I'll see if I can find a link to that.

Here's where you can download the multicore version of Prime. Makes life much easier.
ftp://mersenne.org/gimps/p95v256.zip

The wife will thank me that I didn't go for the Small FFTs on her laptop this morning.



On a side note I did notice this about what RealTemp reports for VID.
When I pull the plug on the laptop the VID reported by CPU-Z and RealTemp drops to a lower value.

[rge]

Thanks for the additional info. I thought there was a separate sensor but I also read that Intel had removed this additional temperature sensor on some of their newer CPUs. I'll see if I can find a link to that.

Some more of intels conflicting documents, or in this case likely a poor editing job. The E8xxx series I think was slated to do away with the cpu diode, as evidenced by page 39, of thermal spec sheet for E8000/E7000 series which states under a graph "Note: The processor has only DTS and no thermal diode. The TCONTROL in the MSR is relevant only to the DTS."
http://download.intel.com/design/pro...nex/318734.pdf

However that document conflicts with spec sheet for E8000 series. And I emailed intel, they did say E8400 has a cpu diode. Also you can use speedfan, go to configure>>advanced>>click on mobo chip, for mine it is It8718F, and cpu temp value is diode. If it were reading mobo socket sensor that value will be thermistor. In fact if you change value to thermistor, it will read bizarre number, showing that it is a diode.

I can only assume that intel was planning on not having a cpu diode on E8xxx series when the document graph was done, then changed their mind, and forgot to change the graph.

[lunadesign]

CompuTronix: I ran into an old copy of your Guide on another site but its definitely not the latest as I don't see any references to Real Temp on it. Where can I get the latest and greatest guide? I've Googled around and all I see are references to deleted items.

BTW, I ran through your calibration procedure and got the following (after calibration) with the case closed and fans running normally:

Tcase = 27-28C idle, 45C load
Tjunction avg = 31C idle (avg excludes 2 cores that don't register this low), 50C load (includes all 4 cores)

Ambient: 23C
Chipset: X48
CPU: Q9450 (stock, no OC)
Cooler: Thermalright Ultra 120 Extreme
Frequency: 2.66GHz
Load: Prime95 - Small FFT's for 10 mins
Motherboard: Gigabyte GA-X48-DQ6
Stepping: C1
Vcore Load = 1.168

Adjustments to SpeedFan:
- CPU: +9
- Cores: -8/-8/-6/-10

Do these values seem reasonable for my configuration?

[lunadesign]

Here's where you can download the multicore version of Prime. Makes life much easier.
ftp://mersenne.org/gimps/p95v256.zip

Thanks! I don't know why I didn't see that the first time....very helpful!

[djstx2003]

hi. i have cpu Intel Core 2 Extreme QX9650 and i have this problem:
the core 1 always stay at 40c and i see in real temp an up and down in temp (min 17c, max 50c) in core 0. also if i run a few times the test sensors always gives me another results. the most times says the core 1=0(stuck) but another times says about core 0 and core 3, the others have value 1.
all this readings is right or something is wrong with the cpu?
here's what it shows me the real temp:



thank you unclewebb for this great program. be well!!!

[alex17 GTX]

What happened with the VID ? Until now both realtemp and coretemp said my cpu(8400 Q746A) has 1.1125 vid. Now both programs says the vid is 1.2125. So what changed?

[Nano74]

Please people , how can i get Real Temp 2,61 ???and some instruction if is not an usual instalation

[Nano74]

sorry guys i did it, but is same sh%&""$·"!!!!, TJMAX 95 for q9450, it stil same as 2,60, , any comments??

[unclewebb]

The VID that most programs are displaying seems to be motherboard dependent and changes to your C1E / SpeedStep settings in the bios or in the Power Options settings in Windows will effect what programs report for VID.

During testing yesterday on a Dell Core2Duo laptop, RealTemp reported 4 different VID values as the mobile chip transitioned from idle to full load.

In my post above:
http://www.xtremesystems.org/forums/...postcount=1409

there are screen shots of two different values and here's a third screen shot during testing last night.



There were 4 different VID values displayed for the same processor in the same motherboard:

1.0625
1.1375
1.2125
1.2750

This motherboard / chip combo also displayed 4 different multiplier transitions.
x6.0 , x8.0, x10.0 , x12.0

Both CPU-Z, when set to report VID, and RealTemp were reporting the exact same numbers for MHz and VID during these transitions. CoreTemp 0.99 also reported these changes but it was reporting slightly different values that were a little lower. CoreTemp reported 0.9500 for the minimum VID and 1.1625 for the maximum.

In my opinion, the way users are comparing VID numbers is meaningless. Everyone assumes that a low VID means that you have a "good", highly overclockable chip but if VID changes depending on the motherboard you're using then whatever all these programs, including RealTemp, is displaying may not mean a hell of a lot. The Intel documentation for reading VID is about as clear as it is for trying to figure out TjMax.

loonym: I think you noticed this VID problem with RealTemp when you ran the same CPU on two different boards. Did both boards have the same maximum VID at full load or were they both different?

Nano74: RealTemp 2.61 is in beta testing right now but I like it's smaller size and likely increased stability better already. Go to TechPowerUp and download version 2.60 and then you can go to my beta section and download version 2.61 and copy the new RealTemp.exe file into your RealTemp folder and replace the version 2.60 exe file.

http://www.fileden.com/files/2008/3/...alTempBeta.zip

I don't like changing the main download until I get some feedback from XS users who have been a great source of ideas and the main reason why RealTemp has become a decent little temp monitoring program.

Edit: Thanks Nano74: for calling my hard work s-h-i-t. What do you know about TjMax that I don't know? If you are convinced that it is some other value than enter that value into RealTemp and be happy. Please post your vast knowledge about TjMax so we can all be enlightened.

[alex17 GTX]

The VID that most programs are displaying seems to be motherboard dependent and changes to your C1E / SpeedStep settings in the bios or in the Power Options settings in Windows will effect what programs report for VID.

Yes that is true. But i have the same settings as before (no c1e) and the reported vid is now 1.2125 instead of 1.1125. This has to do with the programs. CPU-z reports 1.184-1.200 vdef and i thought maybe the board sets the wrong voltage by default. I guess the vid reading was wrong.

[WoZZeR999]

If both programs are reporting different, then it was probably something at the hardware level.

[alex17 GTX]

Realtemp : Displays CPU voltage identification (VID).
Coretemp : Fix: Incorrect VID detection on 45nm desktop Intel parts.

I believe coretemp was reading wrong.

[swiftex]

Realtemp 2.60
Q9450

core0 48
core1 39
core2 39
core3 44

is the above normal or should I re seat hsf?

[lunadesign]

Realtemp 2.60
Q9450

core0 48
core1 39
core2 39
core3 44

is the above normal or should I re seat hsf?

Are those idle temps? Have you applied any calibration factors to Real Temp or is this Real Temp straight-out-of-the-box?

[swiftex]

Are those idle temps? Have you applied any calibration factors to Real Temp or is this Real Temp straight-out-of-the-box?

idle at 3.37ghz
gaming hits 51-59c depending on core
straight outta da box
do not see calibration settings

[WoZZeR999]

What cooling do you have?

[swiftex]

What cooling do you have?

tr ultra 120
rampage formula
8gb mushkin

[lunadesign]

idle at 3.37ghz
gaming hits 51-59c depending on core
straight outta da box
do not see calibration settings

Your idle temps don't seem that bad. I'm running at the stock voltage/frequency and my cores at idle (with no calibration) are at 38/34/40/40 but the last 2 don't ever register below 40 so ignore them.

[swiftex]

Your idle temps don't seem that bad. I'm running at the stock voltage/frequency and my cores at idle (with no calibration) are at 38/34/40/40 but the last 2 don't ever register below 40 so ignore them.

thanks for the response!