Real Temp - New temp program for Intel Core processors

[phill]

Does the RealTemp TestSensors feature show zero movement on all cores?

If your processor ever over heats, catches on fire and melts into your motherboard because of these sensors, then you'll have something to complain to Intel about but you probably have a better chance of winning the lottery than that ever happening.

Yes to the first part of the question I do wonder whether or not it actually matters as I'm not using high voltages for 24.7 operation and even though I'll be crunching on it as long as the PC is on for, I cant see it really causing me a problem.. I only wish I could get Prime stable at 4Ghz as at the moment as core 1 seems to stop within a 1 minute at the moment with 1.4vcore (bios) but the rest seem fine...

I'll let you know how it goes with Scan as they are pretty good when it comes to things not working etc..

[unclewebb]

Heres my Q6600 GO under load.

I've been trying to convince myself that perhaps the IHS is not making good contact with both cores. That might explain why one of the dual cores in a Quad is running at a higher temperature. When the two center cores are running hotter or if the two outside cores are hotter you could consider that it might be a heatsink installation or thermal paste issue. The way your processor is, it sure looks like a different TjMax issue as crazy as that sounds to people that still believe these sensors are perfect.

There's exactly 5C difference in your Max Temps.

My original plan was to only include adjustable TjMax for the two dual cores within a Quad but now I'll just go with adjustable TjMax for each core for more flexibility.

Could you e-mail me a RealTempLog.txt file for your processor. My address is in the About... box.

You can go into the RealTemp.ini file and set:

LogInterval=1

to create lots of data for me. Run it at idle for a minute and then run Prime small FFTs for a couple of minutes until temps more or less stabilize and then go back to idle for a minute so I'll have plenty of numbers to look at.

Information like this really helps me out. The 45nm Quad guys with seriously bunged up sensors will thank you!

[Nevrsadie]

I've been trying to convince myself that perhaps the IHS is not making good contact with both cores. That might explain why one of the dual cores in a Quad is running at a higher temperature. When the two center cores are running hotter or if the two outside cores are hotter you could consider that it might be a heatsink installation or thermal paste issue. The way your processor is, it sure looks like a different TjMax issue as crazy as that sounds to people that still believe these sensors are perfect.

There's exactly 5C difference in your Max Temps.

My original plan was to only include adjustable TjMax for the two dual cores within a Quad but now I'll just go with adjustable TjMax for each core for more flexibility.

Could you e-mail me a RealTempLog.txt file for your processor. My address is in the About... box.

You can go into the RealTemp.ini file and set:

LogInterval=1

to create lots of data for me. Run it at idle for a minute and then run Prime small FFTs for a couple of minutes until temps more or less stabilize and then go back to idle for a minute so I'll have plenty of numbers to look at.

Information like this really helps me out. The 45nm Quad guys with seriously bunged up sensors will thank you!

Well, problem is that I ordered a new hsf, Zigmatek HDT S-1283 to replace my Artic Cooling Freezer 7 pro, because I was not happy with my temps. I received it today after posting my above temps. I am now priming at the same speed and vcore as earlier with new hsf. My temps are better, as expected. But the cores are now reacting like you stated earlier, so I think that thermal grease and position of hsf definately have an effect on the temps of each core. Here is my current temps with exact same bios settings but new hsf.

[Nevrsadie]

Couldnt find your email adress, so I just pm'd you the results. Started idle, then load with prime, then back to idle.

[Ace-a-Rue]

Ace-a-Rue: Thanks for the info. If you have time could you try running this version of Prime and use Small FFTs:

ftp://mersenne.org/gimps/p95v256.zip

I find the load with this test is very steady and keeps the max temps steady as well.

ok....going to bring me and my system to its knees

EDIT:...ok, after several reductions in FSB, i was able to keep all four cores engage...this is 3.9 Ghz for 10 minutes.

[unclewebb]

Nevrsadie: Your original data combined with your new data shoots a big hole in the different TjMax theory. This 5C difference that some of us are seeing under load looks more like a coincidence and a heatsink / thermal paste installation issue. I'll have to pull my heatsink off tomorrow and do some more testing along with a re-mount to see if anything changes.

I finally finished writing some code to handle this problem and now I find out it might not be a problem after all.

Oh well, at least this feature is available if anyone wants to experiment or has their own theories that need to be tested.

Version 2.3 is available for download. It has been uploaded so if you have trouble downloading it then remember to clear your cache because it might be getting an older version form there.

It includes adjustable idle calibration and adjustable TjMax for each core so a user has full control. Adjusting TjMax on a per core basis can make the temp data look believable but if the real problem is in the installation or is caused by the flatness of the IHS or heatsink then you should avoid making this adjustment.

I've seen a lot of RealTemp screen shots and found that very few users were actually using the Idle Calibration feature so I removed it from the GUI. This feature is still available but you will need to go into the RealTemp.ini file to make adjustments. If you do use this feature, it is sort of a set it and forget it thing so I thought it was just extra clutter on the main screen.



The new RealTemp.ini file looks like this:

[RealTemp]
User=0
TjMax0=0
TjMax1=0
TjMax2=0
TjMax3=0
Idle0=0
Idle1=0
Idle2=0
Idle3=0
LogInterval=5
LogFile=0


TjMax values can range from -3 to 3 and will bump TjMax up or down in steps of 5 degrees.

Idle values can range from -2 to 2. Previous Idle Calibration values of (+) or (++) would now be entered into the .ini file as 1 or 2 and a (-) or (--) calibration is equivalent to -1 or -2.

The LogInterval can now contain any factor of 60. (1, 2, 3, 4, 5, 6, 10, 12, 15, 20, 30, 60)
Set this to 60 and it will output a line to the log file once every 60 seconds.

I reworked the timing of the XS Bench feature so if you overclock after you boot up using SetFSB or ClockGen, it should be reflected by a lower time and a higher score.

Instead of clicking the mouse in the TjMax box at the top of the screen to show additional data, now you click outside of the box to the right of TjMax to show the additional data and a mouse click on the left side of the processor name box will reset your Min and Max temp data. I like hiding features for the hard core types that actually read all this stuff.

Here's the usual link.

http://www.fileden.com/files/2008/3/...7/RealTemp.zip

Not that anyone actually reads it but maybe tomorrow I'll update the documentation in post#1. There were a lot of minor things cleaned up and worked over so if there are any problems, let me know.

More testing and then it will be time to start working on that Minimize to System Tray feature so everyone will finally be happy. The NHL playoffs starting tomorrow night might get in the way of that though.

[IanB]

I like hiding features for the hard core types that actually read all this stuff.

As a programmer, I believe that a clear user interface is the most important thng about a program, since if you don't understand what it's telling you or how to use it easily, then the results it displays can be less than meaningful.

Until you made this alteration, I was very impressed with both your concept and your app. But hiding things rather like Easter eggs just for a bit of fun, so they aren't easily accessible, is plain daft. Sorry.

This program is hard enough for people to understand how it works without concealing layers of information behind invisible buttons, without any sort of UI clue that the feature is there. Putting it in the help file just doesn't cut it, I'm afraid.

[steve30x]

This program reports temperatures are lower by about 10 - 15 celcius than Core temp but its the same value higher than speedfan. Also when I Overclocked my CPU to 3.5ghz with 1.475V it read my temperatures on 2 cores to be 70 celcius. I dont know how accurate this is but those temps at full load are realy high. When the CPU was on IDLE the temperature showed 44 celcius.

Also I dont understand how to calibrate this program to my CPU. I know its a bit dumb of me but why cant this program calibrate itself?

PS. Im using water cooling for my CPU

[emoners]

@uncle: the option to calibrate "--/++" is no longer available is the 2.3v...is this a good thing?

[insaneoctane]

@uncle: the option to calibrate "--/++" is no longer available is the 2.3v...is this a good thing?

Dude, look 3 messages above your post, he explains he removed it form the GUI but it is still available in the .ini file

[Ace-a-Rue]

As a programmer, I believe that a clear user interface is the most important thng about a program, since if you don't understand what it's telling you or how to use it easily, then the results it displays can be less than meaningful.

Until you made this alteration, I was very impressed with both your concept and your app. But hiding things rather like Easter eggs just for a bit of fun, so they aren't easily accessible, is plain daft. Sorry.

This program is hard enough for people to understand how it works without concealing layers of information behind invisible buttons, without any sort of UI clue that the feature is there. Putting it in the help file just doesn't cut it, I'm afraid.

i am not totally in agreement with your approach to have everything visible.

the way uncle has changed his program to allow adjustments to the individual cores for both idle and TJmax makes his interface a little more uncluttered...

i think the "minimum temperature" category could be excluded since most of us are really not consummed by minimum temperatures, just maximum!.

lets give uncle some kudos for him designing this program and then being open minded to receive input to improve the ergonomics of his program.

we, as a society, are way to critical of others who put their sweat and toil into a project that he or she is trying to make for the betterment of everyone...where is the appreciation factor!

[jas420221]

i am not totally in agreement with your approach to have everything visible.

the way uncle has changed his program to allow adjustments to the individual cores for both idle and TJmax makes his interface a little more uncluttered...

i think the "minimum temperature" category could be excluded since most of us are really not consummed by minimum temperatures, just maximum!.

lets give uncle some kudos for him designing this program and then being open minded to receive input to improve the ergonomics of his program.

we, as a society, are way to critical of others who put their sweat and toil into a project that he or she is trying to make for the betterment of everyone...where is the appreciation factor!

I think he has gotten PLENTY of kudos in this thread. I do not believe that IanB's post warranted this follow up either. We are all making suggestions and he is doing what is best for all of us. I mean, do you want everyone to give unclewebb a reacharound before they make a suggestion???

[loonym]

Very nice work!!

GO BRUINS!!

[unclewebb]

IanB: You are 100% right. It was late last night, I was tired and I think it was mostly a cheap shot at all the users who don't bother reading anything about the program before running to the forum and screaming, "WTF???"

RealTemp is all about user input. I was able to add a couple of buttons without making the interface too ugly while making it much easier for new users.



I still don't plan to add the Idle Calibration feature back to the main interface though. Of course, I've had my rubber arm twisted before. To me, this is a more advanced feature that is not even being used by most users and if you're interested in using it then you're going to need to read post #1 to really understand the logic behind it. I think a user at this level should be able to go in and do a quick edit of a .ini file. As a user I far prefer editing a .ini file than having junk added to my registry that never gets removed. RealTemp remains 100% self contained in one folder.

As always, I'm open to all suggestions. Version 2.3 is going to have a very short life span and the new and improved version 2.4 will be available by closing time tonight, or when they drop the first puck, whatever comes first.

[TL1000S]

As a programmer, I believe that a clear user interface is the most important thng about a program, since if you don't understand what it's telling you or how to use it easily, then the results it displays can be less than meaningful.

Until you made this alteration, I was very impressed with both your concept and your app. But hiding things rather like Easter eggs just for a bit of fun, so they aren't easily accessible, is plain daft. Sorry.

This program is hard enough for people to understand how it works without concealing layers of information behind invisible buttons, without any sort of UI clue that the feature is there. Putting it in the help file just doesn't cut it, I'm afraid.

"As a programmer.." ??
You are talking about GUI/application design here. Not programming.
Programming is a skill at different levels/different technologies (tools/languages..).
Has really little to do with the layout/design of eg an app like RealTemp.
I am impressed that a *Real* programmer (pun intended) as Unclewebb is able to make a nice looking app as he certainly have to code "low-level".

Apart from the above I agree that a GUI should be intiutive and easy for "novice" users.
But this app is for "Real Geeks".

[Ace-a-Rue]

I do not believe that IanB's post warranted this follow up either.

your opinion...but i disagree with you TOO!

We are all making suggestions and he is doing what is best for all of us.

NO KIDDING!

[unclewebb]

Any user interface suggestions for 2.4? Is the new proposed design more intuitive? For the truly anal amongst us I also made it the exact same length as CPU-Z so it will look good side by side in screen shots with the gold standard of CPU monitoring software.

steve30x: Also I dont understand how to calibrate this program to my CPU.

Did you read through post #1 where I did my best to explain what you need to do? I don't mind helping you as long as you've read the directions first. What don't you understand? The calibration process simply involves comparing your lowest reported idle temperature to your room temperature. The program will still be very accurate at reporting your full load temps even if you don't calibrate it. The calibration option is for users that want or need accurate idle temperatures as well as full load temperatures.

I know its a bit dumb of me but why cant this program calibrate itself?

I'm just trying to mop up the mess that Intel created by:
a) not calibrating these sensors on the assembly line for accurate idle temperatures out of the box.
b) and not providing adequate documentation to the public.

[jas420221]

The usual request... tray temps!!!

[soundx98]

Be nice to be able to hide the additional core readings on my Wolfdale if there is nothing to read.
Fantastic work. Kudos to ya.

[jason4207]

As suggested previously, a different font/bold for the core temp would be nice.

[Ovidiu]

Hello! Ok, am I missing something? Shouldn't the idle temperature at 2000 be much smaller than the one at 3600? Or it will matter only in the load where there's a difference of about 7-8 degrees. Otherwise, congratulations for this little application unclewebb!

[unclewebb]

RealTemp 2.4 with less cryptic features is now available.
http://www.fileden.com/files/2008/3/...7/RealTemp.zip



How's that for listening and responding to user input?

Ovidiu: There are a lot of variables that effect idle temperatures but surprisingly, MHz is not a big factor. Heat is proportional to the square of voltage so your core voltage setting will have a little more effect than MHz but at idle still not a huge amount. Like you said, the biggest changes are at full load when overclocking and especially when using extra core voltage.

Edit: I just noticed that you are using an E8400. Many of these have issues with sensors that get stuck at low temperatures. If you can find a cooler place to temporarily run your computer, you will be able to see if you have this problem at 32C.

[bryanw1995]

I've been trying to convince myself that perhaps the IHS is not making good contact with both cores. That might explain why one of the dual cores in a Quad is running at a higher temperature. When the two center cores are running hotter or if the two outside cores are hotter you could consider that it might be a heatsink installation or thermal paste issue. The way your processor is, it sure looks like a different TjMax issue as crazy as that sounds to people that still believe these sensors are perfect.

There's exactly 5C difference in your Max Temps.

My original plan was to only include adjustable TjMax for the two dual cores within a Quad but now I'll just go with adjustable TjMax for each core for more flexibility.

Could you e-mail me a RealTempLog.txt file for your processor. My address is in the About... box.

You can go into the RealTemp.ini file and set:

LogInterval=1

to create lots of data for me. Run it at idle for a minute and then run Prime small FFTs for a couple of minutes until temps more or less stabilize and then go back to idle for a minute so I'll have plenty of numbers to look at.

Information like this really helps me out. The 45nm Quad guys with seriously bunged up sensors will thank you!

unclewebb, I have a Q9450 on my ip35pro at home, with another one on the way that will go in my ip35e. Do you need additional info on 45nm cpus? I'm not overly concerned with my temps,etc, especially after reading this thread (core temp shows 71/65/64/65, real temp 61/55/54/55 under 24/7 seti@home) but I would be happy to give you some data if you think it will be useful.

edit: temps are at 1.38 vcore in bios @ 8x450, 1.27 vdroop under load.

[unclewebb]

Hi bryanw1995, welcome to XS.

I'd be interested in seeing any idle data for one of these 45nm Quads. If you can run your cpu at low MHz like 266x6 and low core voltage like 1.10 volts and show me your idle core temp as well as your approximate room temperature, that would be great. Run your cpu fan at max and try to get your cpu as cool as possible. Using the above settings helps me make a fair comparison to the testing I've done.

[bryanw1995]

Unfortunately I have an ip35pro, it won't allow me to go below 1.2vcore in bios :mad; I will run it at 1.2vcore and 266x6 with my tuniq on full blast, however. temps had better be VERY low at those settings!!!

btw, this is actually an X3350. I do have a Q9450 that is supposed to be waiting on me when I get home sunday, however. That one is getting stuck in an ip35e, however, and might not be installed for another week or so. I'll try to get you readings from both computers asap.