Real Temp - New temp program for Intel Core processors

Did you look at the file permissions on the RealTemp.ini file? Other users have reported problems when moving the RealTemp folder and this file from one volume to a different volume.

Both TJMax and the calibration factors get saved correctly and also get reloaded correctly for me so I'm not sure what's going on. Make sure the RealTemp.ini file is not open in another program like Notepad or Wordpad while you are saving your settings or while RealTemp is starting up. I'll have a look at my code to see if I can see any problems that might happen with Atom CPUs.

In the Settings Windows you have to click on OK for your new settings to be saved.

I checked file permissions, full control is granted. Tried the new version you recommended with the same results.

Here's the procedure I've employed for changes: I open Real Temp click settings button enter custom value, click apply, click ok and I'm able to observe the change reflected in the temperature reading. However if I close Real Temp and restart it I'm back to square one except that if I browse the ini file I can see my custom values saved there.

Since the changes are written to the ini file I don't believe this problem is related to file permissions(changes to the RealTemp.ini are persistent). What's puzzling is that other changes such as those to font type and color are reflected in the program's operation where as temperature calibration is not. Perhaps RealTemp is ignoring the TJmax and idle offset values in RealTemp.ini in favor of default values in MSR?

Nightstar: Thanks for taking the time to clearly detail this problem. Unfortunately I don't have a solution for this. I don't understand why values are being written correctly to the INI file and other values are being read from the INI file correctly but the TJMax and Calibration factors are being ignored. This doesn't happen on my desktop or laptop computer and after hundreds of thousands of downloads, this is the first I've heard of this problem. If I have some time this weekend, I'll try to add some debugging code to RealTemp so maybe we can find out what's happening to the calibration data.

Thanks for taking the time to investigate this issue. I thought I might have been overlooking some setting or procedure. I figured I should report this issue in case it was a bug.

I seem to recall another fellow posted the same problem in this very forum which was never resolved. Indeed that's what prompted this exchange. If I find the post I'll link it here.

Regardless, It's a nifty little tool, even if I have to set the values each start. Thanks again.

Quote:

---

Originally Posted by unclewebb

RealTemp 3.67
http://www.mediafire.com/?jibd769x6z7n0c8

---

unclewebb, the mediafire link pionts to RealTemp 3.64, not 3.67 (or at least that is what the zip file is called). Is this correct?

Hello Unclewebb

I've got the exact same error as Nightstar. Running Windows 7 64bit, i5-2500k. Its saves the ini file, but do not load the data when restarting the program.

Nevertheless, thank you for a great program! :)

Is it possible to create a realtemp with ROG style colors and icon ? :) same as cpu-z rog
Attachment 117548

I agree that the CPU-Z and GPU-Z Republic of Gamers editions look great but I don't have any plans to do the same for RealTemp.

http://www.techpowerup.com/downloads...OG%20Skin.html

where exactly is 3.67?

There is a link to RealTemp 3.67 in the very first post in this thread. I also counted 4 links to a download that contains RealTemp 3.67 on this page. It's not too hard to find. :)

I've been testing a Gulftown (i7 970) and while I'm no expert I think Intel are playing fast and loose with the TJ Max. Real Temp reports it on my system as 101 however when testing the 970 I noticed it was throttling at a reported temperature of roughly 60ºC. Intel claims that the TCase for this CPU is 67.8ºC and my understanding is that throttling 'kicks in' at around 20ºC above that.

So assuming my 970 isn't defective then for it to reach the temperature when it will begin to throttle i.e. around 88ºC the TCase is in fact 130 not 101. I must admit being surprised at some of the reported lowish temperatures for the Gulftown series and I suppose an easy way to give that impression is for Intel to 'adjust' the TJ Max or am I misunderstanding something?

That is one limitation with Intel writing TJMax information to each core of a Core i CPU. Intel can write whatever they want to that register now and all software will read that register and report the same temperatures. The temperature sensors that Intel uses in these 32nm CPUs are not that accurate. There is an undocumented amount of error in terms of how they set TJMax and there is temperature slope error where the amount of error increases, the further you get away from the calibration point. When combined, the amount of error can be significant.

What sort of throttling are you seeing? When overclocking it is likely that you are seeing throttling based on power consumption. The amount of turbo boost you get depends on whether the CPU is operating within its power envelope. If it goes over the limit, it turns off turbo boost. When near the limit, it rapidly cycles turbo boost on and off and RealTemp will report that as a reduced multiplier. Most motherboards have an option in the bios so you can increase these turbo TDP / TDC power limits so you will continue to get full turbo boost even when overclocking and fully loaded with LinX. ThrottleStop lets you adjust these limits from within Windows but it is designed for the 4 core CPUs so I'm not sure if it works on 6 core CPUs.

Intel CPUs are not designed to throttle based on the TCase temperature. TCase is only intended to be used so computer manufacturers can design a heatsink and fan and a computer case. The theory is that if you design a system and the TCase temperature does not exceed the TCase specification then that CPU will rarely if ever trigger thermal throttling at TJMax. Thermal throttling is controlled by the core temperature which RealTemp reports and that's it. If you ever trigger any thermal throttling, ThrottleStop will change from OK to LOG in the thermal status area to show that the CPU core logged at least one thermal throttling incident since you booted up. If it still shows OK then the CPU didn't thermal throttle.

Some motherboards use the Bi-Directional PROCHOT feature. This can be used to tell the CPU to throttle based on an outside event. For an example, many laptops can throttle the CPU based on the GPU core temperature. I think the voltage regulator can trigger this type of throttling on desktop motherboards.

Post some pics at full load so I can have a look. The i7 Turbo GT program in the RealTemp download is a very accurate tool that can monitor for the slightest hint of throttling.

http://www.mediafire.com/?jibd769x6z7n0c8

Edit: I just noticed your sig. The Asus P6T series has known throttling issues when pushed extremely hard. The special 0006 bios was successful at disabling the throttling when using 4 core CPUs but I'm not sure how that works when running a 6 core CPU. Here's some more info about that problem.

http://www.xtremesystems.org/forums/...-Thread/page74

Thanks for the update. I'm using an Asus P6TD Deluxe and it allows me to turn off turbo and use the extra multiplier for overclocking. I've previously run a Core i7 920 on this motherboard and never seen this behaviour.

The settings I'm using that initiate 'throttling' (not sure what to call it now after your explanation) are a BCLK of 180, multiplier of 25 to give 4.5GHz with a vcore of 1.3875v. I'll do some screen shots but basically what happens is that with Prime95 small FFTs the multiplier drops in CPU-Z from 25 to 24 within a couple of seconds and stays there and the CPU speed is shown as 4.32GHz. Although this in theory should reduce the temperatures they still stay a fraction higher than they do at 4.425GHz (BCLK 177 x Multi 25 at 1.35v). That could be explained though by the slightly increased vcore. At 4.425GHz there's no sign of the multiplier dropping even after a few hours and recorded temperature don't exceed 37ºC above ambient, typically around 20ºC at the time of testing.

Edit: Just seen your edit. I'll spend a bit of time working through the implications of loading that BIOS and depending on the risk might give it a try.

Also forgot to add that I've never had the 'Thermal Status' show anything other than OK. I was aware from your guidance that it would show LOG or PROCHOT if limits had been exceeded.

Before running Prime95

Attachment 118202

A few seconds after Prime95 started and the multi drops to 24

Attachment 118203

The default multiplier for a Core i7-970 is 24.0. Anytime the multiplier goes higher than 24, it is using the Turbo Boost feature of the processor. The Asus bios can be misleading. It might let you disable turbo boost and select a 25 multiplier but as soon as you boot up, in the background it turns turbo boost back on because that is the only way you can access the 25 multiplier in this CPU.

CPU-Z rounds off its data and is not precise enough to show turbo throttling so don't bother relying on what it is telling you.

6 core CPUs draw 50% more power than 4 core CPUs at full load so they hit the turbo TDP limit sooner and will start to throttle back the amount of turbo boost you get. This throttling is not the traditional thermal throttling. RealTemp shows that all of your cores are running OK.

To reduce this turbo throttling you need to increase the TDP limit. You can try using ThrottleStop to do this. One user in that thread reported that he had some success doing this.

ThrottleStop 3.00
http://www.techinferno.com/downloads/?did=1

Your CPU is not an Extreme CPU so the Turbo TDP / TDC power limit values might be locked and not adjustable. Your only solution then would be a different bios. Hopefully that thread I posted will help you out.

Unfortunately Throttlestop has most of the options including the important ones grayed out.

I'm reading up on flashing using the 0006 BIOS. I've not used the Afudos option for years so need to work out how to do it again. I suppose if I use the 24 multi and increase the BCLK then I shouldn't see this throttling although I already know from earlier testing than even a small increase in the BCLK requires an inordinate amount of tweaking.

I'm curious how the TDP limit is calculated as the power draw on the system is only about 35 watts more than with my i7 920 at it's maximum overclock.

Thanks for your help. Even if the BIOS doesn't work at least my concerns about the CPU overheating have been allayed (to some extent).

A Core i7-920 typically started to throttle on these boards when overclocked to about 4000 MHz depending on how much voltage you need to run that speed. An extra 35 watts might not sound like much but it is enough to put you way over the built in TDP turbo power limit. You won't get any turbo boost when overclocking like this and will be left running at the default 24 multiplier and that's it. If the TDP is locked then the only option is the modified bios but I'm not sure how much success people have had when overclocking 6 core non-Extreme CPUs on this board. At least several users have reported success flashing the P6TD to the 0006 bios and if it doesn't work out, you can always flash back to what you are running now using the same method. Good luck.

Just a quick update. I increased the BCLK to 188, dropped the multi to 24 and added a bit to the vcore, QPI/VTT and CPU PLL to give 4.5GHz. Although I only ran P95 small FFTS for 5 minutes there was no sign of throttling and temperatures were OK. I know it will probably need tweaking to get it stable but I ran my 920 at a 200 BCLK with a 21 multi for 18 months so I reckon there's a bit more in it even without the extra multiplier.

I'm really grateful for your advice.

You won't have to worry about any turbo throttling when using the default 24 multiplier because you are not using turbo boost. The 25 multiplier requires turbo boost but not the 24 multi.

On the Asus P6TD, using the default multiplier is the best way to avoid a sagging multiplier and the turbo throttling issues.

RealTemp 3.69.1
http://www.mediafire.com/?4uixpjtezznuzkd

• Fixed number of log file headings when using Core i Dual Core CPUs.
• Moved GPU column to far right end of the log file.
• Improved formatting and added more white space to the ThrottleStopLog.txt file.

any updates?

Updates for what? Do you have any problems? :)

Quote:

---

Originally Posted by burebista

Updates for what? Do you have any problems? :)

---

Has support for the 3960x been added yet??? :) Real Temp GT 3.64 the proc is correctly identified. It only reads temps on 4 cores. 2 are grayed out. :( The regular version isn't working for 6 cores either.

Well then it looks like Kevin has some work to do. :D

Monstru came to the rescue and provided me with some test results from his new 3960X CPU.

As soon as he has a chance to test the new version of RealTemp GT, I will post a download link here.

Quote:

---

Originally Posted by unclewebb

Monstru came to the rescue and provided me with some test results from his new 3960X CPU.

As soon as he has a chance to test the new version of RealTemp GT, I will post a download link here.

---

Woot! Thanks unclewebb! :up: