RealTemp General Discussion

[Growltiger]

In Windows 7 you have to tell the OS to always show all icons for this RealTemp feature to work correctly. I haven't played with Windows 8 recently to see if anything has changed.

Windows 8 works exactly the same as Windows 7. But the icon has not gone into the collection you can't see, it has gome altogether. You can prove this by pressing Customise which shows all the icons. There are only two listed at this moment, and those are both visible. There should be 3.

I just pressed Redraw Icons and now there are 3 visible and 6 listed! More and more confusing.

I set them all to Show, and I got 4. I exited them all. There are 2 left in the list, both shown as Show, but they are not visible.

I'll reboot and start again!

 

[Growltiger]

It is working normally now.

I think there isn't a bug. It is just that Windows decides to hide some icons. One simply needs to tell it to show them.

Sorry for the false alarm.

 

[unclewebb]

Starting with Windows 7, Microsoft assumes that a program will only add 1 system tray icon. If a single program tries to add multiple tray icons, it seems to cause problems for the Windows "Hide the Icons" feature. I guess we will have to wait for Windows 9 for them to fix this feature.

 

[Y3k]

I haven't had any success in getting RealTemp to run at all. I've downloaded v3.70, unzipped it and have run it as administrator by right clicking and selecting run as administrator. I keep getting a window labelled Driver Error stating that Driver Not Found On Network. The WinRing0 libraries extracted to the same directory as RealTemp and I haven't touched anything. I haven't been able to find an answer through Google. I'm running Windows 7 64-Bit on an I7-3770K. Help greatly appreciated.

Ok, I registered to this forum just to answer your question: I had the same error, and I figured out that I was launching RealTemp from a network location (because I'm in a windows domain and I have roaming profiles/folder redirection).

The easy fix is to copy your RealTemp folder in C:\ (or C:\program files(x86) if you prefer of course) and launch the .exe from there.

 

[CAT-THE-FIFTH]

I am having a slight issue here!! I installed a Xeon E3 1220(basically a Core i5 2400 with no IGP and a lower TDP) in my system this week and seem to get conflicting temperature readings.

Real Temp and Core Temp agree with each other whereas HWMonitor reports about 8C higher

I looked at the Tj. Max values and Real Temp uses 94C and HWMonitor uses 102C.

Which is correct??

 

[unclewebb]

Core Temp and RealTemp read the TJ Max value from a register within the CPU. If you downloaded a fresh copy of RealTemp and it shows TJ Max is 94C and Core Temp also shows the same then I would go with that.

MSR 0x1A2 MSR_TEMPERATURE_TARGET
bits[23:16] The minimum temperature at which PROCHOT# will be asserted.
The value is degree C.

MSR Tool
http://www.mediafire.com/?myjkxzkzzmd

If you don't trust RealTemp and Core Temp then you can read the Temperature Target directly. Download my MSR Tool and enter 0x1A2 in the MSR Number box at the bottom and then click on the Read MSR button.

In the EAX box there should be 8 digits. Ignore the last 4 digits to the right. The 2 digits to the left of the last 4 digits is where the Temperature Target info is hiding. Can you post a screen shot of that?

I think you will see the digits 5E. The number 5 and then the letter E.
5E in hexadecimal is equivalent to 94 decimal.

If this is the correct value then you can go into the HWMonitorW.ini file and use that value instead of 102 so all programs will be in agreement.

 

[CAT-THE-FIFTH]

Thank you very much! Someone suggested deleting the hwmonitorw.ini file and restarting it. This seems to have fixed it and now it uses 94C.

 

[Spider-Vice]

Core Temp and RealTemp read the TJ Max value from a register within the CPU. If you downloaded a fresh copy of RealTemp and it shows TJ Max is 94C and Core Temp also shows the same then I would go with that.

MSR 0x1A2 MSR_TEMPERATURE_TARGET
bits[23:16] The minimum temperature at which PROCHOT# will be asserted.
The value is degree C.

MSR Tool
http://www.mediafire.com/?myjkxzkzzmd

If you don't trust RealTemp and Core Temp then you can read the Temperature Target directly. Download my MSR Tool and enter 0x1A2 in the MSR Number box at the bottom and then click on the Read MSR button.

In the EAX box there should be 8 digits. Ignore the last 4 digits to the right. The 2 digits to the left of the last 4 digits is where the Temperature Target info is hiding. Can you post a screen shot of that?

I think you will see the digits 5E. The number 5 and then the letter E.
5E in hexadecimal is equivalent to 94 decimal.

If this is the correct value then you can go into the HWMonitorW.ini file and use that value instead of 102 so all programs will be in agreement.

Hello, I'm sorry to bump this thread, my CPU is a Q6600 G0 stepping, I've been trying to calibrate its core temperatures, but with no success, but before I went into further details by doing sensor tests in RT, I used that tool you posted, and it tells me (by converting the hex to decimal) that my core 0 and 1 Target temperatures are 212ºC?! I might be doing something wrong, I'll leave a screen of that it outputs. From some investigation, I seem to have seen that the Target temperatures of Core 0 and 1 are different from the ones of core 2 and 3... In the documentation on the RT site it says that core 0 and 1 are supposed to be balanced on the sensor test... In my case it's 2 and 3. Weird, but moving on, let me post here the MSR tool screen... Or my CPU has gone nuts or the MSR number is wrong, for which I'd like an answer to which is the Q6600's G0 Thanks.

https://dl.dropbox.com/u/14596765/cpu.png

 

[unclewebb]

The TEMPERATURE_TARGET MSR only applies to the newer Core i CPUs. There is no known similar MSR in the previous generation Core 2 CPUs.

I think the Q6600-G0 is 100C for the first 2 cores. The second set of cores seem to have a slightly higher TJ Max, maybe about 105C for the few chips that I have seen. If you run a consistent load like Prime95 small FFTs and get your CPU nice and hot, you might notice this difference. You usually need to get the core temperature up over 70C.

Intel doesn't publicly document any of this so this is just my best guess based on some hands on testing of a Q6600 GO.

 

[Spider-Vice]

Oops, okay then, thanks. Anyway the sensor tests seem to show a bigger discrepancy between cores 0 and 1 and not 2 and 3 like you say it should be. Cores 2 and 3 seem more balanced in the distance to TjMax, as below (Tested with C1E and EIST off):

This is with stock clocks and stock cooler. Not sure how to interpret those results in order to get more "accurate" core temperatures.

 

[Merk]

Question on i5-3570K temp fluctuation on one core

Hey Unclewebb, I have this question. I think that I may have a faulty cpu.

At idle - low load, I see temps around 35 30 39 26 on my i5-3570K. It is on a Z77X-UP4 TH motherboard and it is not overclocked. Settings are pretty much at default except that I set the RAM to XMP profile. The ambient temperature in my room is at 24C, so that means the 4th core of this cpu would be 2 degrees warmer than the room, which I find hard to believe.

If that's not enough, the minimum temperature if left to idle will end up like this: 29 25 29 19.

Again, a little hard to swallow. 19C would be a full 5 degrees cooler than my room temp and I am not cooling this thing with liquid nitrogen.

Another bit of info, if you watch the 4th core, it will jump from about 24 to 34 in an instant while the other cores remain stable at 34. This happens at idle. It seems like the 2nd core may exhibit similar behavior, but not to the extent of the 4th.

Under load, I get these maximum temps. I am loading it with Handbrake, encoding a dvd to mp4 for iPhone. After about 5 minutes, these are the max temps: 63 61 62 56.

Should I worry about this at all? I can return the cpu to a local Microcenter for an exchange. I don't want to end up with a chip that runs hotter or something worse though.

Thanks for any advise.

 

[unclewebb]

Should I worry about this at all?

No.

I can return the cpu to a local Microcenter for an exchange.

There is no need to return your CPU. The sensors that Intel installs in their CPUs are not 100% accurate temperature monitoring devices. They never have been and it is unlikely that they ever will be.

The purpose of these sensors is to control thermal throttling and thermal shutdown at approximately 105C and 130C. This is where these sensors are calibrated to be most accurate but even here they are still only accurate to +/- 5C. I am 99.99% sure that your sensors are accurate enough to control what Intel designed these sensors to control.

Many of these sensors tend to read low when the CPU is lightly loaded or idle. Your peak core temperature when idle is probably very close to the actual core temperature but without some sophisticated test equipment and a controlled lab, we will never know how accurate it is. This peak value is referred to as the CPU Package temperature and can be read directly from the recent CPUs. The T|I version of RealTemp displays this value since it tends to be a little more meaningful.

RealTemp T|I Edition
http://www.overclock.net/t/1330144/realtemp-t-i-edition

This is a one off version of RealTemp for my friends at Tech|Inferno. I was looking to get some feedback from Ivy Bridge owners. I haven't heard anything too negative so I have to assume that it works OK.

I plan to do a few minor changes for RealTemp 3.80 which will be similar to above and should be ready in the near future. Hopefully the next version can do another million downloads from TechPowerUp like version 3.70 just completed.

 

[Merk]

No.



There is no need to return your CPU. The sensors that Intel installs in their CPUs are not 100% accurate temperature monitoring devices. They never have been and it is unlikely that they ever will be.

The purpose of these sensors is to control thermal throttling and thermal shutdown at approximately 105C and 130C. This is where these sensors are calibrated to be most accurate but even here they are still only accurate to +/- 5C. I am 99.99% sure that your sensors are accurate enough to control what Intel designed these sensors to control.

Many of these sensors tend to read low when the CPU is lightly loaded or idle. Your peak core temperature when idle is probably very close to the actual core temperature but without some sophisticated test equipment and a controlled lab, we will never know how accurate it is. This peak value is referred to as the CPU Package temperature and can be read directly from the recent CPUs. The T|I version of RealTemp displays this value since it tends to be a little more meaningful.

RealTemp T|I Edition
http://www.overclock.net/t/1330144/realtemp-t-i-edition

This is a one off version of RealTemp for my friends at Tech|Inferno. I was looking to get some feedback from Ivy Bridge owners. I haven't heard anything too negative so I have to assume that it works OK.

I plan to do a few minor changes for RealTemp 3.80 which will be similar to above and should be ready in the near future. Hopefully the next version can do another million downloads from TechPowerUp like version 3.70 just completed.

http://img405.imageshack.us/img405/8687/millione.png

Thanks for the quick reply. I'll give the T|I a try and let you know how it goes. After I posted, I started scrolling back through this thread and I read a reply from several weeks ago that explained things pretty well.

 

[Merk]

RealTemp T|I Results

So here is what I came up with using RealTemp T|I. I ran Prime95 blend mode for just under 4 hours. Please see the attached image.

Anyone care to comment on these temps? Ambient air in the room is 24 C.

System is:
i5-3570K stock default clocked 3.4 GHz with 3.8 GHz Turbo
Cooler Master Hyper 212 Evo with one fan in push configuration
Gigabyte GA-Z77X-UP4 TH Motherboard
Corsair Vengeance 8 GB DDR3 @ 1600 MHz XMP Profile
EVGA GTX 260 Graphics Card

The case is an old Thermaltake Damier Xaser V with 3 intake fans (Front, Top and Side), 1 internal blowing toward RAM from empty 5.25" bays, and 1 rear exhaust fan. The fans are not running anywhere near their max rpms. I have them set where the noise is not too loud.

I had attempted a mild overclock (4.4 GHz) using just the Turbo multiplier and 1.20 V. Last night I made it crash when I ran Handbrake for the first time. I realized then that it was no where near stable. Can anyone suggest a proper guide, thread, etc. for learning how to overclock my setup. I had gotten really good with my Q6600 G0 just before I bought this new chip. Now I feel like I have to relearn my whole BIOS.

Thanks again unclewebb for your explanation of the sensors and the purpose of them. I appreciate you taking the time to make me feel more comfortable with Ivy Bridge.

 

[burebista]

First your temps are fine for an Ivy.
Second here is an excellent OC guide written by an respectable OC-er. The guide is not for Ivy CPU's but you'll get an idea.
Or here an Ivy OC guide for a Gigabyte MB.

 

[Merk]

First your temps are fine for an Ivy.
Second here is an excellent OC guide written by an respectable OC-er. The guide is not for Ivy CPU's but you'll get an idea.
Or here an Ivy OC guide for a Gigabyte MB.

Thanks for the links and the reassurance. After I posted those stock clocked temps with the tech inferno RealTemp, I boosted my turbo multipliers on all cores to 40. I ran Prime95 for 30 minutes and had no errors. I put Memtest on an USB drive and ran that for about 14 hours with absolutely no errors.

I always forget about all the little BS battles that you have to deal with when working with a new motherboard. We get things figured out after a while and then we don't think about all those hours spent trying to figure out a little driver issue or whatever. Then one day you decide, ah hell the i5-3570K is down to $169, seems like a good time to buy and there you are all over again, trying to sort out some BS problem. My Logitech G15 (old blue one) had stopped allowing my to hit delete to enter the BIOS. I just figured out that it is because I plugged in the USB hub on my U2410 monitor to one of the USB 3.0 ports on the motherboard. If I plug it into one of the USB 2.0 ports on a front panel connector, the G15 allows me to use it at post. It's one thing after another. I'll save that for another forum though.

 

[Merk]

Just thought I would share the last bit of info with you guys that have been nice enough to help out.

I found this site http://www.overclock.net/t/1291703/ivy-bridge-overclocking-guide-asus-motherboards and basically I used the information I found to adjust the settings on my motherboard.

I started at 43x multiplier and 1.200V. I had to reset my CMOS a couple times, but I ever so slowly made it up to 46x and 1.305V. I had been running Prime95 for about 15 minutes when my CPU alarm went off. Prime95 had passed the first section of tests and when it hit the next section it must have shifted gears because my hottest core had went up from 91 to 102.

I immediately stopped the test and I also immediately knew that I had found my wall. So now I am back down to 45x and I'm working on a stable Vcore. I had run 10 minutes of Prime95 at 1.235V earlier. Temps topped out at 80C for me then. I tried to run it again and failed, so I am now at 1.240V and testing. I'll keep going up .005 until I can get at least 6 hours out of it.

If anyone has a suggestion on making the 46x multiplier work, I'm open to suggestions. My CPU PLL is at 1.7V, VTT is 1.05, VCCSA is .925, and again Vcore was at 1.305 when I almost touched the sun

Capture3.jpg is when I was under 100% load, just before the critical event!
Capture4.jpg is right after I had to pull out the fire extinguisher.

 

[trickeh2k]

Just thought I would share the last bit of info with you guys that have been nice enough to help out.

I found this site http://www.overclock.net/t/1291703/ivy-bridge-overclocking-guide-asus-motherboards and basically I used the information I found to adjust the settings on my motherboard.

I started at 43x multiplier and 1.200V. I had to reset my CMOS a couple times, but I ever so slowly made it up to 46x and 1.305V. I had been running Prime95 for about 15 minutes when my CPU alarm went off. Prime95 had passed the first section of tests and when it hit the next section it must have shifted gears because my hottest core had went up from 91 to 102.

I immediately stopped the test and I also immediately knew that I had found my wall. So now I am back down to 45x and I'm working on a stable Vcore. I had run 10 minutes of Prime95 at 1.235V earlier. Temps topped out at 80C for me then. I tried to run it again and failed, so I am now at 1.240V and testing. I'll keep going up .005 until I can get at least 6 hours out of it.

If anyone has a suggestion on making the 46x multiplier work, I'm open to suggestions. My CPU PLL is at 1.7V, VTT is 1.05, VCCSA is .925, and again Vcore was at 1.305 when I almost touched the sun

Capture3.jpg is when I was under 100% load, just before the critical event!
Capture4.jpg is right after I had to pull out the fire extinguisher.

Either, you're unlucky with a very hot i5 or you're not using an aftermarket cooler. I'm in the phase of OC:ing mine and but haven't managed to get it stable at 4,6 yet. This is what my temps where after 7 hours roughly (4th core failed after that :/) http://dl.dropbox.com/u/63927959/stressvärden.PNG

I have another question. I can't make out really what's right and what's not...
http://dl.dropbox.com/u/63927959/volt.PNG

Why does CPU-Z report a vcore at 1.344v while RealTemp says 1.5012 VID whatever that is? I was kinda shocked when I saw that since I was experementing with turbo offset rather than vcore offset.

 

[Merk]

Either, you're unlucky with a very hot i5 or you're not using an aftermarket cooler. I'm in the phase of OC:ing mine and but haven't managed to get it stable at 4,6 yet. This is what my temps where after 7 hours roughly (4th core failed after that :/) http://dl.dropbox.com/u/63927959/stressvärden.PNG

I have another question. I can't make out really what's right and what's not...
http://dl.dropbox.com/u/63927959/volt.PNG

Why does CPU-Z report a vcore at 1.344v while RealTemp says 1.5012 VID whatever that is? I was kinda shocked when I saw that since I was experementing with turbo offset rather than vcore offset.

I am using a Hyper 212 Evo.

I had set my Prime95 up with "Custom" settings by checking the bottom box when starting the stress test. The minimum fft size is 8kb this way (it may always be, not sure) and the second pass is when it uses the 8kb. I guess the way it works, the smaller the fft size the hotter your chip gets.

I haven't tried much more with my OC. I am sitting at 1.250V and 45x multiplier. All my other voltages are the same as above.

 

[unclewebb]

VID is the amount of voltage the CPU is requesting. When you go into the bios and manually set the core voltage, the motherboard ignores the VID signal and goes with the voltage you selected in the bios. In theory, low VID CPUs usually overclock better than a high VID CPU. That's the reason I included VID information in RealTemp.

Merk: I thought I was the only guy that liked punishing my CPU with 100C core temperatures but it looks like you beat me by 1C.

http://img27.imageshack.us/img27/6216/torturetest.png

I have found that Ivy Bridge runs very reliably even at some extreme core temperatures. In the pic I was abusing both the Intel CPU and Intel GPU but neither one complained.

 

[mlee49]

Hi UW, any thoughts on adding an OSD option?

I came across a thread where someone wanted a monitoring program that also could display the temps while gaming. Sadly I couldn't recommend my favorite temperature monitoring program, but I thought I would at least ask if its possible.

Also, CONGRATS ON A MILLION DOWNLOADS!!!!!

 

[unclewebb]

mlee49: I wish I had more time for project RealTemp but unfortunately I don't. There are lots of features that I would have liked to add to RealTemp but unless I win the lottery, that's not likely to happen. These days it takes me a year just to do a few minor updates.

 

[spoidz]

RealTemp changing CPU ratio of overclock when activated?

Hello Uncleweb,

I asked about some calibration temp issues a while back in the main page but never got any replies, so my sensor cal may still be off.

Decided to overclock my I7-960 to start playing Skyrim and I'm getting a weird behavior when trying to use RealTemp to monitor stress testing.

I can get straight into 4Ghz Win7 Ultimate no problems and CPUZ shows good overclock at 160x25. All good.

But as soon as I activate RealTemp at 4GHz, CPUZ shows the ratio drops to 24 and the overclock drops back to 3.8Ghz.

I already ran some Prime 95 on the initial 3.8Ghz and it was OK temperatures wise before I tried to move up to 4Ghz.

Is this a known glitch on either RealTemp or the P6X58D Premium? Or RealTemp in conjunction with CPUZ. It's like RealTemp is hitting the MB to read and corrupting/changing that ratio from BIOS.

First two attach is the Sensor Calibration but I'm not sure if I have the CPU set right according to instructions. Does stock MHz mean normal 24 ratio for the I7-960 CPU or always lowered to 12 per instructions? My ambient at the PC intakes is ~24C. Are we to read ambient inside the case instead?


Third is a good 4GHz overclock into Win7-64 Ultimate with no issues.

Last two is what happens as soon as I turn on RealTemp at 4GHz.

After a couple seconds RealTemp changes that ratio back to 24 every time.

Thanks for any help.

 

[unclewebb]

You should run the sensor test with your CPU at its normal speed with the 24 or 25 multiplier. My best guess for your CPU is to use TJ Max settings of 100, 104, 100, 104 for your 4 cores. This will make the temps line up a lot better from idle to full load. Without a lab and some expensive test equipment, I have no way to prove if using these offset TJ Max settings will result in temperatures that are more accurate or less accurate. At least they will look nice! That's why I quit trying to calibrate sensors. It was a good idea at the time but there are too many unknowns so on some CPUs, it ends up being too much of a guessing game. I have done zero hands on testing with the first generation Core i CPUs so take my advice with a grain of salt.

As for your other problem, I might have an answer for that. Have a look in the RealTemp - Settings window. There is an option in there called Disable Turbo. Is that box checked? If it is, that would explain why running RealTemp is disabling your Turbo Boost and dropping your multiplier to the default which is 24.

The ability to disable Turbo Boost was a useful feature for laptop owners but sometimes this feature gets accidentally checked. I am not sure if this is a RealTemp bug or just user error. The options available in the bios can be a little misleading. To access the 25 multiplier in a Core i7-960, the Intel Turbo Boost feature has to be turned on. Some bios versions used to give users an option to run the 25 multiplier and to Disable Turbo Boost at the same time. After you click on Save in the bios, it analyzes your choices and if it sees that you would like to use the 25 multiplier then it secretly turns Turbo Boost back on regardless of what you selected for this option. That ends up being very confusing to users. They believe Turbo Boost has been disabled when the bios enabled it in the background before booting up.

Anyway, long story. For the next version of RealTemp, I have decided to make it more difficult for Disable Turbo to be accidentally checked. Let me know if clearing the Disable Turbo box fixes your problem.

Do you know about the 26 multiplier that is hiding in your CPU? It's only available when a single core is active and is only available when the C3 or C6 low power states are enabled. As soon as a second core becomes active, it will drop back down to a maximum of 25.

 

[spoidz]

Thanks,

Yes the Turbo switch was the problem.

So I set your recommended TJMax. Do we need to bother then with Idle calibration at all?

My idles at even 3.8Ghz sit at 40's. My Ambient in the room is almost constant 23-24C and 26-27C inside bottom of case.

That's Delta well above what your notes expect, especially on a Noctua NH-D14.

Anyway thanks again for the help