Real Temp - New temp program for Intel Core processors

[unclewebb]

My theory for all Intel Core C2D or Core i7 CPUs is to run them as hot as you like as long as they're stable.

TjMax means they will start to throttle when they get close to this number (usually 98C). Though not documented, the shutdown temperatue of Core 2 Duo Desktop processors is somewhere between 125C and 130C. Those are all big numbers but if Intel thought you were going to hurt your CPU by running it at those temperatures then they would lower TJMax and the thermal shut down point to prevent a pile of warranty claims. They obviously have a lot of confidence in the products they sell.

When overclocking C2D, I've always found that you will need to leave yourself more temperature headroom from TJMax the more you overclock to maintain stability. From what I've seen so far, Core i7 is similar but you can run them reliably a little hotter. I think when Kyle at [H]ard OCP was testing, he was losing stability and getting blue screens when the temps were around 90C. I think that's about 20C hotter than most Core 2 Duo chips like to run at when well overclocked.

Is that a more thorough answer and does it tell you what you need to know?

http://enthusiast.hardocp.com/articl...50aHVzaWFzdA==

[bowman]

I'm running WCG now, been running it for hours and some cores have been in the 87-89 range.. No issues so far.

[Demo]

@unclewebb,

Thank you for the speedy reply.

[rge]

Wife and I did early Xmas...here is my i940. Only had up for few hours, so no clue what I am doing yet. 4ghz stable will be easy and my temp sensors seem to be accurate. speedfan hasnt quite got temp sensors down yet, reporting my load temps at 34C and weird reading.

Realtemp seems to be working well on mine for temps! My sensors read from 33C up, no sticking yet. Here is my test, mine seem accurate. When I get real wb in week or two, will IR this thing and compare to tjmax temps. I screwed up the idle test, thought it was finished, it would read 58C...but I was running prime through half of it.

This is on "water", crooked, homemade ugly bracket half** holding rotated old dtek in "place".

[unclewebb]

rge: You guys are making me jealous.
4GHz isn't too shabby for someone that doesn't know what they're doing yet.

speedfan hasnt quite got temp sensors down yet, reporting my load temps at 34C and weird reading.

RealTemp was similarly borked early in the week. Glad I bust my balls all week getting it ready for you.

With that chip I don't think I'd be thinking about taking the IR gun to it, yet. At Ultra Idle type settings it might be easier to control without a heatsink on it because I think the idle cores get turned off and power consumption and heat are reduced.

I might need a helper later this weekend to use my MSR Tool to see if they can find anything that looks like VID info hiding in there. How about a quicky XS Bench to see how Core i7 compares to Core 2 Duo. I know one thread of XS Bench is not what Core i7 is all about but I still like seeing the numbers! Have you run a Super PI in the 9s yet?

CPU-Z 1.48.4 Beta

[rge]

I definitely appreciate you getting realtemp working for these so quickly And so far its telling me I need a real 1366 wb.

One quick bench before I hit sack...had marathon night, only 3 hrs sleep getting this built, reloading software, water etc...

superpi sub 10 is with my same so far 24/7 setting...this is still 22x182 4ghz, just turbo is kicked in with 23 multi. Tomorrow will bench for high mhz, maybe outside until I get better cooling.

[unclewebb]

Nice numbers after a day of playing with it. Looks like core temperatures are going to be the limiting factor when overclocking Core i7.

One thing I tried to fix earlier this week was this long time problem.
Here's a pic borrowed from Coolaler's site:



If anyone knows if this is fixed then let me know. The degree C sign has been messed up since day 1 for a lot of users outside of North America. I think the RealTemp Thermal Status feature might come in handy with Core i7.

[Dunderhoken]

Max Temp - 45c

Roomtemp - 21c

[Tutto]

Where can I get the latest version of Real Temp?

[unclewebb]

Permanent RealTemp beta link:
http://www.fileden.com/files/2008/3/...alTempBeta.zip

Version 2.86 is the current beta.

Dunderhoken: I would use TJMax=100C for your QX9650. I realize that Intel said at the IDF that the QX9650 has a TJ Target of 95C but I think for accurate temperatures you would be better off if you used TJMax=100C.

[Demo]

Hmm, my temps have been increasing over the past 8 months and i recently cleaned my TRUE-120x and Fan ( open case ) i've been searching my older posts for temp reference seems that i was getting 55-57c before now i seem to be getting 67-68 normally and 72c on very hot afternoons..

[unclewebb]

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

Just a couple of minor bug fixes to the Clock Modulation feature in the Settings Window. 0.0% is supposed to be an undefined state but a few of the Core i7 CPU screen shots were showing that number so I did some fixing.

VID information used to be stored in several different model specific register (MSR) locations but that information has disappeared in Core i7. With rge's help, I'm trying to find one register that contains this information. I haven't found any Intel documentation yet about how to extract this information so this is just a wild guess.

Edit: Locating VID info was not successful so I've blocked that for Core i7 users in version 2.88 which is now available. Without some help from Intel, I don't like my chances of figuring this out.

Demo: What CPU are you using. I know RealTemp bumped up TJMax by 5C for many CPUs. Is that part of the increase? If voltage and MHz haven't changed then that's hard to explain what's going on. The CPUs I've had have been pretty consistent.

[AKHandyman]

Unclewebb ... I wanted to tell you that I ended up with the Xeon back in the Gigabyte mobo. I took the QX9770 out and I am waiting on a ASUS Rampage Formula to arrive. In the meantime, I did run the test on the Xeon, but was interrupted and never could finish. I didn't get the screenie. But do you know if it is normal to see a 6~10 degree difference between the two cores? (Core #0-#1, and Core#2-#3) Just curious and I will post one tomorrow for you to see. It's a little late for me to crank that rig up.

[T_Flight]

The VID numbers listed for the Core i7 965 on the Intel Proc Spec Finder were 0.85-1.3625v

CPUID String - was listed as 106A4h

Thermal Spec - was not listed

Thermal Design Power (TDP) - was 130w

[unclewebb]

Should you be seeing a 6C to 10C difference? Probably not. Is it normal? Unfortunately it does seem to be the norm.

Intel has talked about slope error but my Cool Down Test is leading me to believe that TJMax isn't nearly as fixed as we all once assumed. There could be as much as 10C of error at TJMax so two similar CPUs or similar cores could report vastly different temperatures. I'm still learning about this stuff so when you get a chance tomorrow post a screen shot of that test. It helps me see what the biggest contributor to the differences in temperatures is.

Thanks T_Flight. There are some nice charts about VID in the documentation but they don't include what register in the CPU you're supposed to read that value from or how you're supposed to interpret it. Anyone with a Core i7 should post a screen shot so I can see if I did good or bad.

Section 2.5 of Volume 1 of the Core i7 Datasheet tells me almost everything I need to know, almost.
http://download.intel.com/design/pro...hts/320834.pdf

[Demo]

@unclewebb

Same everything for the last 8 months, i actually forgot if those temps i posted before were already at 100C TjMAX or still at 95TjMax, well at 65c right now im at 35 away to TjMax, anything above 25 to TjMax is presumed safe right ?

[T_Flight]

Should you be seeing a 6C to 10C difference? Probably not. Is it normal? Unfortunately it does seem to be the norm.

Intel has talked about slope error but my Cool Down Test is leading me to believe that TJMax isn't nearly as fixed as we all once assumed. There could be as much as 10C of error at TJMax so two similar CPUs or similar cores could report vastly different temperatures. I'm still learning about this stuff so when you get a chance tomorrow post a screen shot of that test. It helps me see what the biggest contributor to the differences in temperatures is.

Thanks T_Flight. There are some nice charts about VID in the documentation but they don't include what register in the CPU you're supposed to read that value from or how you're supposed to interpret it. Anyone with a Core i7 should post a screen shot so I can see if I did good or bad.

Section 2.5 of Volume 1 of the Core i7 Datasheet tells me almost everything I need to know, almost.
http://download.intel.com/design/pro...hts/320834.pdf

Will post screens when i get mine up and running. I've run into a few roadblocks one of which is my burner died and the new one in my sig might get here wednesday...maybe. If not it might be thursaday.

I will post screens of that as soon as I can get the dive all setup and get it running. I'm at that phase now where this thing is slowly killing me. It on the table setup, but needs a couple parts, and then when i go to slipstream my Win install I find my burner is crapped out.

The good news is everything will be neew and fully compliant though.

[unclewebb]

Demo: Thermal throttling usually happens when the Distance to TJMax is down to about 2 or 3. In my opinion, as long as your CPU is not throttling then it's perfectly safe. The problem I found when overclocking Core 2 Duo is that when your CPU is near its maximum 24/7 overclock, you won't be able to run Prime stable anywhere near TJMax. Your Distance to TJMax might need to be 30 or even 40 to maintain Prime stability. My air cooled E8400 needs closer to 40 when I'm using 1.40 volts to remain stable but at default MHz and voltage I can run Prime right up to TJMax.

Every CPU is unique so I hate to draw a line in the sand and say, "Don't go past here, or else." If you're Prime stable and not throttling, you don't need to worry about temperatures.

T_Flight: Good luck with your build. Nothing more frustrating than having to wait for that one last part. I look forward to seeing some more Core i7 screen shots.

Edit: I just found out that VID for Core i7 is still broken. This info that is located in several places within Core 2 Duo chips has disappeared from Core i7. I'll keep working trying to find where this information has gone to.

Edit #2: I wanted to back up what I said before about thermal throttling so here's a good example.

At 4 GHz, my E8400 can't run at these temperatures but by backing my overclock down to 3.6 GHz, I had no stability issues while running Prime95 Small FFTs for 3 hours with my Tuniq fan turned off.



Only once did it get hot enough that it actually hit TJMax at 100C which you can see in the SpeedFan graph. The rest of the time it would intermittently throttle when hitting a Distance to TJMax of 2. This cycled the multi down to 6.0 (2400 MHz) which kept the temps from getting too out of control.

I've always thought the CPU sensor reads a little high on this board and with it showing 107C, that kind of confirms it.

At 4 GHz, I think Core i7 has more temperature head room than a Core 2 Quad at 4 GHz. The bottom line is that if you're stable and not throttling, you really don't have to worry about your core temperature.

[JohnZS]

Hi Unclewebb

I braved a bit of a small overclock.

QX9650 @ 3.4Ghz 1600Mhz FSB. In order to obtain PRIME and Intelburntest Stability I had to raise my V-Core to 1.25V. (stock is 1.12V).

Sensors are now moving.....at different rates?



Seems at stock V-Core the sensors move uniformly (all 9) yet with raised V-Core they become more independent.

Have you seen this before?

Or is this another example of how bizarre my sensors are?

Thanks

Regards

John

[unclewebb]

JohnZS, you're just the guy I like seeing data from. My goal is to turn your random number generator into some meaningful temperatures and I think I'm very close to that goal.

This new Cool Down Test is really helping me see what's going on. I took the Intel IDF presentations at face value but all they initially did was get me off track. Intel's graph of slope error was more of a generic line on a piece of paper than the engineering type data I was hoping to see.

With my test, when checking for slope error, I like to compare the 87.5% level to the 12.5% level. I shortened the heat up phase of my test so users wouldn't have to wait too long for the results. For that reason, the 100% level is not your maximum Prime95 temperature but it's usually pretty close to it. The 100% level is also a maximum temperature where 87.5% and lower are averaged temperatures. That's why I like to start by looking at the 87.5% level first.

Anyhow, when comparing 87.5% to 12.5% on your CPU I get the following temperature deltas:

Core 0 = 16.5°C
Core 1 = 15.3°C
Core 2 = 16.1°C
Core 3 = 16.4°C

This shows that slope error is only playing a very minor part in your wildly looking random temperatures. The majority of error in your sensors is at TJMax. The biggest problem is that each core has a slightly different TJMax.

Intel publicly stated at IDF that there is error at TJMax but has only been willing to define that number as plus or minus X for the 45nm Dual and Quad Core CPUs.

Here's the published spec for the 45nm Intel Atom N270 sensors as listed in the datasheet:
5.3 Digital Thermal Sensor
The digital thermal sensor (DTS) accuracy is in the order of -5°C ~ +10°C around 90°C; it deteriorates to ±10°C at 50°C.

If this spec has nothing to do with the 45nm Dual and Quad core CPUs then in their IDF presentation they should have defined exactly what this spec is. They admitted that there is some error at TJMax but have only been willing to refer to the amount of error as 'X'.

Based on the Atom spec and the data coming from sensors like yours, I think assuming that X is somewhere around 5°C is reasonable. That means the Intel calibration point that TJMax is based off of might only be accurate to plus or minus 5°C.

The values released at the IDF conference are referred to as TJ Targets. TJMax is defined as always being equal to or above that Target value. In this case, TJMax would be more accurately defined as 105°C ± 5°C or somewhere between 100°C and 110°C. That range is a lot closer to the truth than the single TJMax=100C number.

That's the problem. Users including myself have always taken TJMax as a very fixed value with virtually no error in that number. The endless argument about what is the correct TJMax has always been flawed. There is no single TJMax number for every QX9650 processor or for any processor. The 65nm sensors may have had a little tighter thermal sensor specs or they might have been able to do a better job of matching similar sensors with each other on the same CPU but there is always going to be some error. Each processor within a given line is always going to have a range of correct TJMax values. Early testing shows the Core i7 is continuing this tradition.

Knowing this, some users are simply going to give up and say, "Accurate temperatures from these sensors simply aren't possible. There are too many unknown variables." That's what Intel has been hinting at since day 1 but the user community has never been willing to accept that.

My opinion is that 100% accurate reported temperatures may not be possible due to the limitations of these sensors but with a simple calibration, and a few tweaks, you're going to get a lot closer to an accurate temperature than what some of these sensors give you out of the box.

rge's recent testing gives us a point of reference at the low end and reading between the lines of what Intel has recently said gives us more flexibility at the top end.

The present RealTemp calibration formula needs to be reworked and users are going to have to be willing to adjust TJMax on a core by core basis in some situations. I'll come up with a new formula and some calibration numbers you can try later this week. I'm hoping for 4 temperature curves that look more like mirror images than random numbers.

My E8400 is listed as 1.225 volts on the box so there's nothing wrong with 1.25 volts. Didn't my previous post show you that these CPUs can take a licking!

[randomizer]

Here's the published spec for the 45nm Intel Atom N270 sensors as listed in the datasheet:
5.3 Digital Thermal Sensor
The digital thermal sensor (DTS) accuracy is in the order of -5°C ~ +10°C around 90°C; it deteriorates to ±10°C at 50°C.

Aren't those sensors more closely related to (but not the same as) the 45nm desktop chips, ie. the ones which we aren't complaining about as much any more? Clearly the DTS is not the same in both 65nm and 45nm chips.

It's somewhat amusing that we've gone from praising the 65nm chips to hating them

[unclewebb]

It's somewhat amusing that we've gone from praising the 65nm chips to hating them

You must have misread my extra long post. The 65nm sensors are definitely better than the sensors used on the 45nm Quad processors. JohnZS has one of the more extreme examples but he's not the only 45nm Quad owner with reported temperatures all over the place.

His CPU doesn't seem to follow the "slope error" theory that Intel has talked about though. Plus or minus half a degree of slope error over a run of 16 degrees is pretty minor. The only other explanation for the vast differences in his temperatures is error at TJMax which Intel fully agrees does exist.

This graph provided by Intel at the IDF helps show the problem:



Intel should have released how big X really is. In JohnZS's case, you can blame a couple of degrees of error on slope error but the other 10 degrees of error is because of differences in TJMax between cores.

Users have never been willing to mess with TJMax but in this case, I think John would have far more accurate temperatures if he did. I plan to come up with some numbers for him to see if we can get those 4 cores tracking a lot more closely then what they are now.

[randomizer]

Perhaps I should have said "hating what Intel is telling us, or isn't telling us, about them." That would have been closer to the intended meaning.

JohnZS's sensors contradict what Intel was saying about the 45nm sensors having better calibration than the 65nm ones because they're all over the place. It makes it all the more likely that the Tj Targets for 65nm are simply not right rather than the sensors just being more poorly calibrated. But why they released numbers that are way too low, and then "fixed" them, when the 45nm ones aren't is just plain strange as I mentioned before.

[elmysterio]

I wanna know the correct TjMAX for : e6850 65nm

[unclewebb]

I wanna know the correct TjMAX for : e6850 65nm

Intel says 90C, RealTemp uses 100C.

The best thing to do is to do your own test. Open you case, crank up your CPU fan it it's adjustable and go into the bios and set your CPU to 266 MHz x 6.0 and set the core voltage to approximately 1.10 volts. Now boot up and compare your reported idle temperatures to your room temperature. Try each TJMax value above and see which one is more believable. Post a screen shot or two of your results.

Intel's IDF presentation originally used 80C for your CPU but when the complaints started coming in, they admitted that it was a mistake and decided to bump it up to 90C. In my opinion, even that is too low. They also call this a TJ Target and admit that the actual TJMax can be higher than the spec that they released. I like 100C better based on IR thermometer testing but I'm interested in what your testing shows.

Users have to get away from the idea that TJMax is some sort of exact value. Intel fully admits that it is not. So far they haven't been willing to release how big a range it might fall into. I'm guessing at plus or minus 5C for the 45nm chips and maybe a little tighter than that for the 65nm sensors.

It makes it all the more likely that the Tj Targets for 65nm are simply not right rather than the sensors just being more poorly calibrated.

That's pretty much my opinion but I'm just one voice against the giant corporation. These processors run and overclock well so the average users isn't too concerned if their reported temperatures are off by a few degrees.