Real Temp - New temp program for Intel Core processors

[DGill]

Not sure about mine.

OCCT shows 99% utilisation
RealTemp 3.11 indicates 75%
& for what its worth, a Vista Gadget also shows 75%

[unclewebb]

Version 3.07 was just for you darkzone but trying to find your missing cores was causing problems for other users so I've gone back to my original code for versions 3.1x and above. There's no difference in features between 3.07 and 3.11 so run whatever version works for you.

The Load meter is accurate on my Q6600 G0.



It should show something like the above when running 4 threads of Prime95 Small FFTs.

This load meter is based on reading timers within the CPU. If you're only getting a reading of about 75% when running the above load then it looks like it's only reading 3 of the 4 cores correctly. Why? I don't know but if you have some time to do some testing I'll come up with a version for you to try to find out what's going on. Just send me a PM and tell me you want to do some testing.

Hardass: If you have RivaTuner installed you could try using the RTCore.dll plug-in. Just go into the RealTemp Settings window and click on the RivaTuner button and tell RealTemp where you have your RivaTuner.exe file located and it will install the plug-in in the correct directory for you. It shows individual core load now so it might help me find / understand where the problem is. The plug-in is burebista approved.

maGSImum: It looks like your sensors are reading a little low at idle which isn't uncommon. Try running Prime95 Small FFTs with some more core voltage to get your temps into a higher range if possible. I found my Q6600 likely has a higher actual TJMax on cores 2 and 3. TJMax = 100 / 100 / 105 / 105 seems more accurate for my CPU. At higher temperatures I have a very consistent 5C difference between the two sets of cores but yours look like your 4 cores would be well balanced at higher temperatures.

The best thing to do is use rge's calibration guide lines. Your sensors are functional at high Distance to TJMax values but you can improve their accuracy by calibrating them.

http://www.xtremesystems.org/forums/...postcount=2429

Add on 1C for a Q6600 compared to his numbers.

DGill: I plan to make a separate Load utility using these timers to do some more testing of this. It's interesting that the Vista gadget is showing the same thing.

[Hardass]

Hardass[/B]: If you have RivaTuner installed you could try using the RTCore.dll plug-in. Just go into the RealTemp Settings window and click on the RivaTuner button and tell RealTemp where you have your RivaTuner.exe file located and it will install the plug-in in the correct directory for you. It shows individual core load now so it might help me find / understand where the problem is. The plug-in is burebista approved.

All 4 cores show 100% in RT

[DGill]

Ahh ... I see. Prime95 works fine, runs 4 cores and RealTemp shows 100% utilisaion. Presumably when I ran OCCT it only identified/used 3 cores.

[unclewebb]

Can you try running Prime 95 Small FFTs? By default, Prime95 does its stress testing at the lowest priority so your computer is still usable while stress testing and usually RealTemp doesn't have a problem reading the correct Load %. Some other load testing programs run at a normal priority which can bog things down. RealTemp might be fighting with some of these stress programs to read the timers correctly.

I've seen this issue before with LinX which is why I boosted the priority of the RealTemp / RivaTuner plug-in above normal to try to correct for this. I might have to add the same code to RealTemp 3.11 if this is the issue.

You might also be able to go into the Task Manager and lower the priority of OCCT to see if that gives RealTemp enough time to read the timers.

[Hardass]

Can you try running Prime 95 Small FFTs? By default, Prime95 does its stress testing at the lowest priority so your computer is still usable while stress testing and usually RealTemp doesn't have a problem reading the correct Load %. Some other load testing programs run at a normal priority which can bog things down. RealTemp might be fighting with some of these stress programs to read the timers correctly.

I've seen this issue before with LinX which is why I boosted the priority of the RealTemp / RivaTuner plug-in above normal to try to correct for this. I might have to add the same code to RealTemp 3.11 if this is the issue.

You might also be able to go into the Task Manager and lower the priority of OCCT to see if that gives RealTemp enough time to read the timers.

Running Prime it reads correctly.

[unclewebb]

Running Prime it reads correctly.

Thanks for that test. I'll try bumping up the priority of RealTemp like I did to the plug-in to give it a better chance of reading the CPU timers correctly when fully loaded with LinX and OCCT and other stress programs that operate at normal or higher priority.

[unclewebb]

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

I boosted the priority one notch to see if that helps maintain the Load accuracy readings while running OCCT, LinX, etc. without RealTemp becoming a CPU hog. Can't have that.

I also put some modified darkzone code in there to try to help him without it screwing up the rest of us.

Place your vote.

Bugs Fixed:
Bugs Created:

[Hardass]

No change while running F@H smp client.

[darkzone]

Nope, still don't work for me Anyway, a fresh install of OS is planned when i7 940 arrives, and I'll see then if the problem is solved.

[RejZoR]

Maybe guys at TechPowerUp could help you.

http://www.techpowerup.com/gpuz/

They're the makers of GPU-Z which reads AMD cards temperature...

[unclewebb]

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

Here's lucky 3.13. I boosted the priority up to the same as what the RealTemp / RivaTuner plug-in uses. In theory, you shouldn't have to ever boost the priority but other programs like OCCT boost the priority of their processes which can prevent RealTemp from accurately reading the CPU sensors. When Prime95 is stressing your CPU it drops its priority level down to rock bottom so other processes can still do their thing. That's the polite thing to do when you are fully loading a CPU. If this fix works, great, if not, I probably won't change it.

If you are still having problems then you can use the Task Manager and right click on RealTemp.exe and set the Priority sky high to see if that makes any difference. I'm curious.

RejZoR: I mentioned that to W1zzard but I don't think he wants to share.

I wouldn't share that code either. This issue has become less of a priority for me since I installed a Nvidia 9800GTX+. I'm happy with RealTemp as is.

[DGill]

RT 3.13 does the job

[unclewebb]

Thanks for confirming that DGill. I'm happy to see that RealTemp is finally working 100% for you or maybe I should say 99.1%.

I think I'll add a darkzone=1 INI switch to the next release so everyone will finally be happy.

[HDCHOPPER]

3.13 rocks

[darkzone]

I think I'll add a darkzone=1 INI switch to the next release so everyone will finally be happy.

Thanks

Finally, i7 940 arrived, in this weekend I'll set up new rig and proceed to a fresh install of Server 2008.

[unclewebb]

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

A few minor bug fixes so that the RealTemp RivaTuner plug-in and RealTemp share the load timers better.
I've also switched to the latest WinRing0 driver. If there are any new bugs, I'll have something new to blame!

[darkzone]

Well, I know you won't like this, but ...

[unclewebb]

Well, I know you won't like this, but ...

Stuff like that used to bother me darkzone but not any more. You've got 3.07 that works great on your OS. I'll try to add that INI file option tonight to see if we can get 3.15 working for your combo as well. It will be interesting when you do a fresh install of Server 2008 this weekend. I'm looking forward to some more bug reports.

[DJSUB]

thanks for the hardwork & keeping us updated with the best temps monitoring program

keep up the good work uncle!

[DJSUB]

I just want to ask if there is a way to change the color of this text?



thanks in advance

[unclewebb]

I just want to ask if there is a way to change the color of this text?

That's a bug in the old C++ compiler I'm using when the rounded modern looking buttons are being used.

I have a demo version of C++ 2008 which might include a fix for this bug and will let me change and control the color of those headings.

I'll put it on the short, things to do list to see what's possible.

[unclewebb]

I decided to switch to Visual Studio 2008 and found that all the easy ways outlined on the net simply don't work when you are using the modern XP/Vista style.

I did find a work around though that doesn't look too bad.



DJSUB: This is probably going to be a custom version for you. Since you're not using an Nvidia card would you like me to get rid of that white 1067A button in the top left corner? I think it will look better that way. I'll just put the 1067A in green with the rest of the data like the good old days.

I'll probably make one of each version. The heading colors will match whatever text color you choose. If someone needs this color customizable then I might throw an INI file text color option in there for you.



On second thought, I might just throw a color picker in there for the headings.

[Hoss331]

Every time I close RT 3.13 and 3.14 I get a windows error "Real Temp has stopped working" but 3.00 closes just fine.

[randomizer]

3.14 closes fine for me.