Real Temp - New temp program for Intel Core processors

[unclewebb]

I'm going to take some images of the differences and do some more testing before I will be satisfied.

Thank God I'm not the only crazy person who appreciates the ability to create your own fonts! Your present label as an "Xtreme Member" is most fitting.

Hopefully I don't figure out how to make anti-aliased fonts show up nicely in the tray area and wreck all of your hard work.

I just spent the last few hours working on getting the Settings window to pop up perfectly on either the left or right side of the main window. I've heard perfection is a sickness and they're probably right!

[Talonman]

We wouldn't have it any other way...

[WizardofOz]

Has anyone ever noticed that the temperature on the Cores change when adjustments in the Bios are made. The value are still the same but are being reported by different Cores, this is especially true for Core's 1,2 and 3. Core 0 is usually constant in reporting the same temp. For Example 33 28 28 31, then at next boot you see 33, 31, 28 28.

Does anyone have an explanation for this? Just Curious as to what is happening.

[thefiddler]

Heya!
Firsts, thanks for this prog!

I just installed my new computer and wanted to start monitoring temps to see if everything is going well. The CPU is an Intel Xeon 3350, mobo Asus Rampage, Thermalright XP120 eXtreme w/ Noctua NH-P12 fan set to its max speed.

I tried the calibration method explained in the documentation (FSB to 266MHz, 1.10v, ratio 6.0 --> cpu at 1600MHz) and launched RealTemp.

Room temparature is 25°C.

The core temperatures reported by the prog at idle were:
41 - 32 - 40 - 42

As room temp is 25c, I guess I should do a "-" calibration to adjust core temps at idle. If I set calibration to -3, 0, -3, -3 I get the following idle temps reported by RealTemp:
38 - 32 - 34 - 38

Does that seem right?


I also ran the Test Sensors function, which reported the following sensor movements:
3 - 3 - 0 - 3
I guess this means the core3 is sticking?

[unclewebb]

WizardofOz: Depending on what you're changing in the bios, you are probably just seeing the natural, somewhat random variation in core temperatures. These temps change instantaneously and depending on how many background tasks you have running and how they are scheduled to each core can change your idle temps by a degree or two. In an open case with a good air cooler and with a minimum of junk running in the background, your idle temps should be pretty consistent. When you do a CTRL+ALT+DELETE, how many tasks are running on your computer at idle?

thefiddler: The numbers you're posting are not consistent. First of all we need to both be talking about the same thing. In a Quad core processor, cores are numbered from 0 to 3. CPU-Z, RealTemp and Intel all use this format. In your case it is possible that core2 is sticking.

You show that your temps with no calibration are 41 - 32 - 40 - 42 and then with a -3.0 calibration you show temps of 38 - 32 - 34 - 38. If a -3.0 calibration factor gets you a 6C drop on core2 then it should also get you the same on core0 but your core0 only dropped 3C. That immediately shows me that there is a problem with your testing method. First thing to check is the same as what I posted above. You need to open up the Task Manager and make sure you have a bare minimum of background processes running. The CPU Usage History graph will show you how much background activity is going on. You also need to idle long enough to make sure your idle temps are consistent. If you are using Vista then getting your computer close to a truly idle state might be a bigger challenge than doing this test in XP.

The best thing to do is to turn on the Logging feature in RealTemp and set the Logging Interval to 1 second. Idle for a minute, run Prime small FFTs on all 4 cores for a minute and then go back to idle for a minute. PM me the information in the RealTempLog.txt file after that so I can get a more accurate look at your data than just a snapshot. This will help me determine if you do have a sticking sensor. You can run the logging test at low volts / low MHz as well as at default volts and MHz. The more data the better.

If you do have one or more sticking sensors then before doing any calibration you need to try and determine which sensors are sticking and at what DTS number they are sticking at. All this is a lot of hassle but we have to work with what Intel has given us to try and get some accurate idle temps out of the on chip sensors that were neither designed or calibrated for that purpose.

[Ming@s]

Another program very useful to see temperatures.

[unclewebb]

Here's how my Q6600 looks when it is fairly idle.



It typically sits at 0% with the occasional spike to about 3%. You need to get your temps as low as possible during testing and the best way to do that is to make sure your computer is not working on some background task. Can someone with Vista post a similar screen shot? Maybe my calibration test isn't suitable for Vista.

[Demo]

Vista x64, left for 5mins idle.

[clokker]

How come it's not showing the number of running processes?

[gmcg]

Here's how my Q6600 looks when it is fairly idle.
It typically sits at 0% with the occasional spike to about 3%. You need to get your temps as low as possible during testing and the best way to do that is to make sure your computer is not working on some background task. Can someone with Vista post a similar screen shot? Maybe my calibration test isn't suitable for Vista.

It is perfectly suitable for Vista 64 Ultimate ... you are the best, uncleweb!
Thanks for your hard work!

Also I would like to add one more picture - it is under load - just started Everest Lavalys Ultimate stress test - Everest, Coretemp, HWmonitor show the same error - randomly (temp jump up to 124C), and your RealTemp is a winner! It definitely shows a REAL TEMP

[unclewebb]

Thanks gmcg. I guess not following the herd was a good idea.

Good software never rests. I've been playing around with a new gamer mode today. It blasts on the screen the temperature of your cores in the upper right hand corner while you're playing. The image isn't the most stable and flickers since it has to be continuously updated but the temp numbers are visible. I don't recommend this option for 24/7 use but I think lots of users are curious when they are in the middle of a game how hard their cores are being worked. My overclocked Q6600 was barely warming up during some 3DMark05/06 benching.

I also added on a Pop-Up Info option so if you want to see all the temps of all your cores over each tray icon then just select this. Otherwise, they will be labeled core1, core2, core3 and the temp info will only pop up over the left most icon the way I like it.

When you click on the Settings button, the Settings window will now open up immediately to the left or the right of the main RealTemp window, where ever there's the most room, so you don't have to drag it over each time. I hate to work too hard! I also equalized the height of the windows so it looks well balanced.

I'll release another beta tomorrow so you can check it all out.

[gmcg]

Thanks gmcg. I guess not following the herd was a good idea.

Good software never rests. I've been playing around with a new gamer mode today. It blasts on the screen the temperature of your cores in the upper right hand corner while you're playing. The image isn't the most stable and flickers since it has to be continuously updated but the temp numbers are visible. I don't recommend this option for 24/7 use but I think lots of users are curious when they are in the middle of a game how hard their cores are being worked. My overclocked Q6600 was barely warming up during some 3DMark05/06 benching.

I also added on a Pop-Up Info option so if you want to see all the temps of all your cores over each tray icon then just select this. Otherwise, they will be labeled core1, core2, core3 and the temp info will only pop up over the left most icon the way I like it.

When you click on the Settings button, the Settings window will now open up immediately to the left or the right of the main RealTemp window, where ever there's the most room, so you don't have to drag it over each time. I hate to work too hard! I also equalized the height of the windows so it looks well balanced.

I'll release another beta tomorrow so you can check it all out.

Thanks, unclewebb, great ideas! It looks that Nvidia BIOS in 790i monitors the same temps as Everest/Coretemp/HWmonitor, not as yours and as soon as we have that error, either it is a sensor error or a driver error, BIOS uses that information and stops/freezes the operations to prevent motherboard/CPU from overheating. Thats why we probably had random freezes in 790i and so many people are upset.
I assume Nvidia should contact you and use your method of thermal monitoring to get the real data and prevent those errors ...

[Demo]

How come it's not showing the number of running processes?

I was also posting the same image on another forum for a different purpose, since ive already edited the image ( resize,removed system uptime and process )

If you need to know my processes are 36 on start up.

[slim142]

I need some help

How do I configure this to correctly read the temps of a Q9450?

[unclewebb]

How do I configure this to correctly read the temps of a Q9450?

Have you tried reading through the documentation yet? Is there anything you don't understand?

http://www.techpowerup.com/realtemp/docs.php

You have to do a little homework first and then explain your problem a little more thoroughly. The sensors on all these processors are unique and behave differently from core to core so no one has any magic calibration numbers that will work for all processors.

gmcg: Interesting info. It's not rocket science getting temperature info out of a Core processor.

RDMSR 0x19C
DTS = ( EAX >> 16 ) & 0x7F

I'll let the engineers figure out what core they're reading and they can guess at TjMax like everyone else is doing to convert this DTS value into something that is close to an absolute temperature.

[colt1911]

Unclewebb, is the 2.58 version a beta or has it been released yet? I googled for it and could only find 2.5 version. Thanks for a great program and all your hrad work keeping it updated for us.

[unclewebb]

Beta version 2.58 is available.

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

Let me know if the Gamer Mode is a usable feature for you. It worked in 3DMark06 when I tried it on my Quad and it only cost me about 10 points. It provides some interesting info when you can read it. If I find a way to make it clearer I will but I don't think there's much you can do when you're trying to overwrite the screen of a game moving at a variable frame rate.

On games with lots of FPS it seems usable and displays in game temps without having to try and do an ALT+TAB which never really works. Are there games that it doesn't work on? Is this experimental feature worth keeping?

Great work uncle or
what the hell were you thinking?

I'm presently working with one user that needed some more negative calibration options for his 45nm chip so I also opened that option up to -4.0 instead of only -3.0.

[Talonman]

Gamer Mode works fine here. Not a bad feature to have, and further sets RealTemp apart from the pack.

[k4vz0024]

Uncleweb, I have a problem with when I run XS Bench. The orograms just frozen. However, If I used Real Temp 2.5 the same test working fine. I have Q6600 and Xp pro.

Tx

[Talonman]

Same under Vista 64 bit. We have us a small bug.

[ZoKi]

nice little program dude thank you

[unclewebb]

Thanks for the feedback. I think I know what I screwed up and it should be easy enough to fix. Later today.

[msgclb]

Using Vista x32 when I run XS Bench Real Temp 2.58 goes into the 'Not Responding' mode. I waited a respectable time before killing it. Real Temp 2.5 and 2.57 passes ok. Both show (Not Responding) while running the test but complete there test.

Looks like I was a little late!

[unclewebb]

Here's a quick fix.

I'm not sure if there was one or two problems. So far I think I've fixed one problem. Version 2.58.1 is available for beta testing.
http://www.fileden.com/files/2008/3/...alTempBeta.zip

Try running an XS Bench without Gamer mode on and see if it works and then click on Gamer Mode and see if that causes a crash. Let me know what OS you're using. XP has no complaints but writing information directly on screen might be a no no in Vista. If it isn't allowed in Vista then I'll have to ax this feature.

msgclb: The Not Responding does happen sometimes during the XS Bench due to this process trying to hog the whole computer so the results are repeatable but obviously 2.58 had a big fat bug. I've changed the way RealTemp calculates MHz recently to try and get the CPU MHz option to recognize changes when SetFSB is used. That part is working well on my board and is just as accurate as CPU-Z but I'll have to wait for some feedback to see if it is working for everyone.

[msgclb]

Here's a quick fix.

I'm not sure if there was one or two problems. So far I think I've fixed one problem. Version 2.58.1 is available for beta testing.

Try running an XS Bench without Gamer mode on and see if it works and then click on Gamer Mode and see if that causes a crash. Let me know what OS you're using. XP has no complaints but writing information directly on screen might be a no no in Vista. If it isn't allowed in Vista then I'll have to ax this feature.

msgclb: The Not Responding does happen sometimes during the XS Bench due to this process trying to hog the whole computer so the results are repeatable but obviously 2.58 had a big fat bug. I've changed the way RealTemp calculates MHz recently to try and get the CPU MHz option to recognize changes when SetFSB is used. That part is working well on my board and is just as accurate as CPU-Z but I'll have to wait for some feedback to see if it is working for everyone.

v2.58.1 XS Bench without Gamer mode on: Pass
v2.58.1 XS Bench with Gamer mode on: Pass

I had to clear my cache again!
I understand the Not Responding!