Real Temp - New temp program for Intel Core processors

[djstx2003]

thanks my friend, i'll do that. i'll put -2.0 to cores 2-3 and leave tjmax at 95. i see about 1-2 degrees more than cores 0-1 but i think its normal because they have different distance to tjmax.

thank you my friend!!! keep the good work!!! be well!!!

regardz

djstx2003

[unclewebb]

Use the same TjMax for all cores, balance the idle temps by adjusting core2/core3 and that's about as good as it gets without having to invest in a fancy IR thermometer. Now go do some overclocking and show us the only numbers that are really important, BIG MHz.

[FatboyHK]

thx for this excellent tool. Almost perfect, but if I can suggest an improvement, I would like to have an alarm function that can trigger other application, for example an shutdown script, instead of just start an alarm. That would be more useful.

[rusty]

once again god bless unclewebb for incredible work!!!

[1badhabit]

Great program, very accurate!! Thanks.

[unclewebb]

thx for this excellent tool. Almost perfect, but if I can suggest an improvement, I would like to have an alarm function that can trigger other application, for example an shutdown script, instead of just start an alarm. That would be more useful.

There's always room for improvement! I'll look into including something like that for the next release.

Thanks everyone for the positive feedback.
Initial testing looks good. 59 hours and still going strong.

[Justintoxicated]

I just got this program and my Q9450. My temps are all over the place. I am liquid cooled but i dont see what difference that would make.

My temperatures are: 37 43 45 45 and when testing the sensors temperature movement is 7,9,3,3 in the same order.

Should I RMA my CPU?

[Xello]

Nope, everyone seems to have whacky temperature readings and as Web has said it comes down to the fact that the sensors on the cpu are far from perfect. It's normal to have readings far apart like that, Justintoxicated, so don't worry.

You can change certain values in the program to sync the cores more but there's a possibility that 1 or more temp sensors are 'stuck', meaning they won't show a value below a certain amount. My core 1 and 2 sensors are like this, they won't go below 34 and 36 respectively. It doesn't mean there's a fault with the CPU though.

If you glance over the last page or 2 you will see some data we provided Uncle with and he was able to advise us on how to set our calibration values, you might want to do the same

[003]

There's always room for improvement! I'll look into including something like that for the next release.

Thanks everyone for the positive feedback.
Initial testing looks good. 59 hours and still going strong.

Another feature suggestion I would have would be to have an option for RealTemp to display the average temp of all your cores in the sys tray instead of one at a time.

[unclewebb]

I just got this program and my Q9450. My temps are all over the place.

Your odd ball temps look pretty normal for a 45nm Quad. Intel neither designs nor calibrates these sensors to accurately report low temperatures. They are calibrated to be extremely accurate at 95C and any usability beyond that is a bonus.

If you read the documentation and try doing the suggested calibration, you might find that with RealTemp they are at least usable even if they are not perfect.

There's not much use doing an RMA because there is no guarantee that your replacement will be any better than what you've got now. Do some research and let me know if you have any questions.

[unclewebb]

It's been mighty quiet in this thread for the last week.
Here's what I've been working on.



I did a quick update to the GUI so the buttons, etc. are XP style. The toggle button in the top right corner is a little wider so it should be easier to click on.

The PROCHOT# section seemed to confuse some so I cleaned it up as well. If your processor is running fine it will display OK. If it is presently throttling it will say HOT and if there was a thermal throttling episode since you started up then it will report that as LOG. I might have to pull the plug on the CPU fan tomorrow to make sure this is working properly! That new IntelBurnTest should get RealTemp lighting up like a Christmas tree.

Edit: IntelBurnTest was causing too many fluctuations in core temperatures so I used Prime small FFTs instead. In the updated pic, core0 is thermal throttling, core1 has thermal throttled but is not currently throttling as shown by the LOG entry and core2 and core3 are still OK and have not done any throttling.

The logging file has been updated to include the date and if you hover the mouse above the Min or Max area it will pop up a date in case you run your computer for days / weeks and can't remember when that happened. Also added a Reset option to the System Tray menu for easier / quicker access.

I'm thinking about allowing users to customize the background and text colors like this which might blend in better with the generally dark Vista themes:



Hopefully a beta of all this good stuff will be ready in a few days for some testing. Everyone must be happy because I haven't heard any complaints lately!

[Shadowthor]

looks good Can't wait for the beta.

[nuclearjock]

unclewebb:

I just went E8500 -> E8600

Do you know if Tj Max has changed with the new E0 stepping??

Do you know what Tj Max should be for the E8600???

[unclewebb]

nuclearjock: I believe that TjMax has not changed and it is still 95C for the E8600. I have not had my hands on one and pointed the IR thermometer at it but I'm pretty confident that TjMax hasn't changed.

X-bit Labs compared an E8600 (ES) engineering sample to an E8500 - C0 and with both processors overclocked to 4275 MHz, the E0 reported temperature was 2C higher based on a TjMax=105C for both of them. They came to the conclusion that this proves that the new stepping must have a different TjMax but I think their understanding of heat output from Core processors as well as the behavior of these sensors is lacking. They also tried to make some direct comparisons of uncorrected idle temperatures which anyone familiar with RealTemp and these sensors knows is a rather meaningless comparison.

http://www.xbitlabs.com/articles/cpu...o-e8600_4.html

After a 10 minute test, a 2C difference proves nothing. A slight change in thermal paste application or the amount of time that it's had to set up could cause that. Even a tiny change in the squareness of the IHS could change temps by a degree or two very easily.

There is also no documentation from Intel that states what the repeatability of these sensors are. A 50C chip might give off a reading of 49C one day and 51C the next day even though the CPU is at the exact same temperature.

After taking every thing into account, my interpretation of the X-bit Labs report is that nothing has changed concerning TjMax. The difference in temps is not statistically significant enough to support their conclusion. A typical error of +/- 1C for both processors would also show a difference of 2C.

[nuclearjock]

nuclearjock: I believe that TjMax has not changed and it is still 95C for the E8600. I have not had my hands on one and pointed the IR thermometer at it but I'm pretty confident that TjMax hasn't changed.

X-bit Labs compared an E8600 (ES) engineering sample to an E8500 - C0 and with both processors overclocked to 4275 MHz, the E0 reported temperature was 2C higher based on a TjMax=105C for both of them. They came to the conclusion that this proves that the new stepping must have a different TjMax but I think their understanding of heat output from Core processors as well as the behavior of these sensors is lacking. They also tried to make some direct comparisons of uncorrected idle temperatures which anyone familiar with RealTemp and these sensors knows is a rather meaningless comparison.

http://www.xbitlabs.com/articles/cpu...o-e8600_4.html

After a 10 minute test, a 2C difference proves nothing. A slight change in thermal paste application or the amount of time that it's had to set up could cause that. Even a tiny change in the squareness of the IHS could change temps by a degree or two very easily.

There is also no documentation from Intel that states what the repeatability of these sensors are. A 50C chip might give off a reading of 49C one day and 51C the next day even though the CPU is at the exact same temperature.

After taking every thing into account, my interpretation of the X-bit Labs report is that nothing has changed concerning TjMax. The difference in temps is not statistically significant enough to support their conclusion. A typical error of +/- 1C for both processors would also show a difference of 2C.

Many thanks....

[WoZZeR999]

It's been mighty quiet in this thread for the last week.
Here's what I've been working on.

I did a quick update to the GUI so the buttons, etc. are XP style. The toggle button in the top right corner is a little wider so it should be easier to click on.

The PROCHOT# section seemed to confuse some so I cleaned it up as well. If your processor is running fine it will display OK. If it is presently throttling it will say HOT and if there was a thermal throttling episode since you started up then it will report that as LOG. I might have to pull the plug on the CPU fan tomorrow to make sure this is working properly! That new IntelBurnTest should get RealTemp lighting up like a Christmas tree.

Edit: IntelBurnTest was causing too many fluctuations in core temperatures so I used Prime small FFTs instead. In the updated pic, core0 is thermal throttling, core1 has thermal throttled but is not currently throttling as shown by the LOG entry and core2 and core3 are still OK and have not done any throttling.

The logging file has been updated to include the date and if you hover the mouse above the Min or Max area it will pop up a date in case you run your computer for days / weeks and can't remember when that happened. Also added a Reset option to the System Tray menu for easier / quicker access.

I'm thinking about allowing users to customize the background and text colors like this which might blend in better with the generally dark Vista themes:

Hopefully a beta of all this good stuff will be ready in a few days for some testing. Everyone must be happy because I haven't heard any complaints lately!

Being skinnable (or customizing colors) would be cool. Maybe allow a background image to be displayed with a separate option for normal gui background and mini background.

Also, IntelBurn (linpack frontend), is not good for a constant heat source. It's good for heating the proc up, and getting it to a maximum high, but prime still wins for a constant heat source. Linpack has to cycle too much data through ram to have constant calcs. It's similar to how blend is cooler than Small FTT.

[unclewebb]

WoZZeR999: Now that RealTemp works pretty good, I'm trying to learn a few things to make it look a little better. The background image idea has crossed my mind, especially for mini-mode. I'm hoping to have separate text and background colors available for both regular and mini-mode.

It took about 15 minutes with no CPU fan to get up to the throttling point for a screen shot. Small FFTs brought the temps up nice and steady. With a decent air cooler, a CPU fan is almost optional!

[Hondo=TFD=]

Which is more accurate? Everest or Real Temp?

[unclewebb]

Which is more accurate? Everest or Real Temp?

Here's some of the real world testing that is behind RealTemp.

http://www.xtremesystems.org/forums/...&postcount=573

All of the competition is basing TjMax on Intel documentation for the mobile processors. I don't believe any of that is relevant to TjMax for the desktop processors.

At the end of the day, you'll have to decide for yourself. You can read the 78 pages in this thread if you're really curious and need some more background info.

[WoZZeR999]

Wow, I didn't realize that there was already 78 pages. Good work uncle!

[Hondo=TFD=]

Here's some of the real world testing that is behind RealTemp.

http://www.xtremesystems.org/forums/...&postcount=573

All of the competition is basing TjMax on Intel documentation for the mobile processors. I don't believe any of that is relevant to TjMax for the desktop processors.

At the end of the day, you'll have to decide for yourself. You can read the 78 pages in this thread if you're really curious and need some more background info.

That's cool. Was just asking, there many different programs.

[unclewebb]

Hondo=TFD=: Try downloading the latest version. It should get your MHz correct, finally! Accurate temps were my first priority. If you take the time to read the documentation and calibrate your sensors, I think you'll be happy with RealTemp and the accuracy of the temperatures it reports.

http://www.techpowerup.com/downloads...Temp_2.70.html

Good work uncle!

Thanks WoZZeR999. It was getting kind of quiet around here so I figured I better create some more features to keep this thread going.

What does a person have to do around here to get sticky status!

[WoodButcher]

What does a person have to do around here to get sticky status!

Maybe a little parlor trick? part the waters? walk over it instead?
http://www.techpowerup.com/downloads...Temp_2.70.html +1 for the sticky!

[Warboy]

You can fix the reading in everest if you take the Tjmax from Realtemp and apply it in options in everest.

[JohnZS]

It's been mighty quiet in this thread for the last week.
Here's what I've been working on.



I did a quick update to the GUI so the buttons, etc. are XP style. The toggle button in the top right corner is a little wider so it should be easier to click on.

The PROCHOT# section seemed to confuse some so I cleaned it up as well. If your processor is running fine it will display OK. If it is presently throttling it will say HOT and if there was a thermal throttling episode since you started up then it will report that as LOG. I might have to pull the plug on the CPU fan tomorrow to make sure this is working properly! That new IntelBurnTest should get RealTemp lighting up like a Christmas tree.

Edit: IntelBurnTest was causing too many fluctuations in core temperatures so I used Prime small FFTs instead. In the updated pic, core0 is thermal throttling, core1 has thermal throttled but is not currently throttling as shown by the LOG entry and core2 and core3 are still OK and have not done any throttling.

The logging file has been updated to include the date and if you hover the mouse above the Min or Max area it will pop up a date in case you run your computer for days / weeks and can't remember when that happened. Also added a Reset option to the System Tray menu for easier / quicker access.

I'm thinking about allowing users to customize the background and text colors like this which might blend in better with the generally dark Vista themes:



Hopefully a beta of all this good stuff will be ready in a few days for some testing. Everyone must be happy because I haven't heard any complaints lately!

Looks good
Just when you think RealTemp can't get any better...it does
Keep up the good work and thanks.
Regards
John