Real Temp - New temp program for Intel Core processors

[Xello]

Xello: I can honestly say that this is the first I've ever heard of this problem and off hand I can't think of anything RealTemp is doing that would cause this. Next time it happens post a screen shot and any details you can think of like if it was Minimized to the system tray or if it was in Mini-Mode or if the RealTemp dialog was just sitting on the desktop and then you had a look at it an hour later and it was black. If you move the mouse back and forth over it will every thing return to normal or do you have to restart? It does sound like a Vista / driver issue. What card are you using and what driver version?

I concede that this is likely a problem elsewhere then i guess. I never leave it on full display, always minimized to tray, and i'll bring it to full display every few hours to check the max temps. It's when i do this that it happens, after several hours of being minimized. I'm doing another prime test right now so i'll take a pic if it happens tonight. Moving my mouse over the window doesn't return it to normal, but the cursor does change from an arrow to a verticle line when i move it over the areas where the fields of text are. I'm on Palit 9600GT with 175.19 WHQL (the latest).

That's a 9600gt issue not realtemp I think. I have occasional weirdness with my 9600's too.

Sounds like that could be it then, though i haven't had anything else like this so far. I'll be getting 4870X2's in a few weeks so i'll see if it does it with them.

Another question, it's probably covered somewhere in this thread but it's a big thread! The 'idle calibration' feature, basically how do i use it to set up properly with my cpu, or is there even a need to, can i just leave it alone?

[WoZZeR999]

The first page has calibration info. So does the TPU page I believe.

[unclewebb]

When running RealTemp if you open up the About... box and click on the TechPowerUp icon it will take you to the main RealTemp info page. Drag that page about half way down and click on Installation & Calibration. Kind of long and wordy but it's a good explanation of what goes on with these sensors sometimes and what you can do to check the accuracy of your sensors at idle where they are usually inaccurate.

Thanks for your detailed explanation of the problem you are having. It will be interesting to see if your problem goes away when you switch video cards. I can't think of anything RealTemp is doing that would cause this. I do know that WHQL display drivers are not always perfect. That's why new versions keep coming out.

[Xello]

Ok thanks i'll check that out.

For some unknown reason print screen key will not function when the realtemp window is open in its 'blackened' state, so i had to use my phone.

Clicky

As you can see i can even click the dialogue boxes and the windows open, but they too are either black or see-through. I'll try to download the latest version, and i'll probably also install the new geforce 177.79 beta drivers sometime today or tomorrow, perhaps that will resolve this anomoly. It's quite annoying as if i restart the program i obviously lose the max temps that were there!

[unclewebb]

Maybe we should call that the new RealTemp Ghost mode. As a programmer, when you don't know how to fix something, you call it a feature. Windows has lots of features!

I'd start by using the latest beta version of RealTemp so I don't have to go looking for a bug that might have already been fixed if RealTemp is to blame for this. Different video card drivers would also be a good idea. I'm hoping when you switch to an ATI card that this problem fixes itself.

Has anyone with Vista x64 and an ATI card ever seen something like this?

[Xello]

:edit:

I had the winring0.dll error but i copied the files from the old folder and it works now.

I tried running the sensor test, i have movements of 14 14 8 and 14, should the 8 be a concern?

[unclewebb]

The Test Sensors applies an equal load to all 4 cores and then compares the core temperature you started at to the core temperature you ended up at. If your 4 cores are equal at idle and 3 of them go up by 14 degrees and one of them only goes up by 8 then that is a sign that something isn't quite right with your core2 sensor. The typical problem is that core2 is getting stuck at idle so even if the core temperature gets cooler, this core will continue to report the same temperature.

To look for this I start by turning on the Log feature of RealTemp and I set it to an interval of 1 second. Let your computer idle for a minute so you can get some baseline data, then run Prime95 small FFTs on all 4 cores for a couple of minutes and then stop that and let your computer idle again for a minute or two. Stop RealTemp and have a look at the log file.

When Prime starts, the core temps should start increasing on all of your cores by a more or less equal amount. When you stop running Prime, the opposite should occur on each core. When a sensor is stuck, as the CPU cools down some of the cores will be decreasing at a similar rate while as your CPU continues to cool, one sensor will stop moving and will continue to report the same temperature even while the other cores are still reporting that they are cooling down.

This isn't a big problem but it's something you need to be aware of. If the sticking point is at 40C then that sensor will never be able to report a temperature less than that. As long as you know that, it's not really a problem. Above the sticking point, most of these sensors work fine and it won't limit the ability of a processor to overclock or anything like that.

Send me a log file of the above test if you're not sure what it's telling you. In CSV format would be nice.

[Xello]

The Test Sensors applies an equal load to all 4 cores and then compares the core temperature you started at to the core temperature you ended up at. If your 4 cores are equal at idle and 3 of them go up by 14 degrees and one of them only goes up by 8 then that is a sign that something isn't quite right with your core2 sensor. The typical problem is that core2 is getting stuck at idle so even if the core temperature gets cooler, this core will continue to report the same temperature.

To look for this I start by turning on the Log feature of RealTemp and I set it to an interval of 1 second. Let your computer idle for a minute so you can get some baseline data, then run Prime95 small FFTs on all 4 cores for a couple of minutes and then stop that and let your computer idle again for a minute or two. Stop RealTemp and have a look at the log file.

When Prime starts, the core temps should start increasing on all of your cores by a more or less equal amount. When you stop running Prime, the opposite should occur on each core. When a sensor is stuck, as the CPU cools down some of the cores will be decreasing at a similar rate while as your CPU continues to cool, one sensor will stop moving and will continue to report the same temperature even while the other cores are still reporting that they are cooling down.

This isn't a big problem but it's something you need to be aware of. If the sticking point is at 40C then that sensor will never be able to report a temperature less than that. As long as you know that, it's not really a problem. Above the sticking point, most of these sensors work fine and it won't limit the ability of a processor to overclock or anything like that.

Send me a log file of the above test if you're not sure what it's telling you. In CSV format would be nice.

Thanks, i'll try that tonight. Generally my idle temps are 32,32,36,25 so i wonder if they should actually be 32,32,32,25

[-X-hellfire]

...it happened again today

did a Registry Mechanic cleanup, on repaired 23 things, maybe 5 was some sort of temp stuff, rebooted and RT still not minimized and takes 30 second to start showing temperatures


started some firefox windows and tabs and then closed them down and had only one firefox window and a tab open when this happened

same as before but now it was firefox RT





...denied it

don´t get it, was is going on

[W1zzard]

realtemp 2.70 posted: http://www.techpowerup.com/downloads...Temp_2.70.html

[unclewebb]

Thanks W1zzard.

-X-hellfire: I just remembered that there is one time when RealTemp will try to connect to the internet. Your warning would only make sense if you opened up the About... box in RealTemp and clicked on the TechPowerUp logo. That would then try to open up your default browser, Firefox, and open up the RealTemp home page at TechPowerUp. That's the only time RealTemp tries to connect to the internet and it doesn't send your latest Visa statement or anything else to TechPowerUp. It should just open up that page and that's it.

Here's the C++ code:

// Close the About window and head to the TechPowerUp website
void CAboutDlg::OnTechbutton() 
{
	ShellExecute( NULL, _T("open"), _T("http://www.techpowerup.com/realtemp/"),NULL, NULL, SW_SHOWNORMAL );
	SendMessage(WM_CLOSE,0,0);
}

I'm not sure about using SendMessage to close the About dialog after someone clicks on it but it seems to work so I left it. If anyone knows a better way to accomplish the same thing then let me know. Maybe I'll eliminate the automatic closing of this window so I won't have to send any of these suspicious messages.

Now if you have a virus on your computer and it is infecting your .exe files like RealTemp then there isn't much I can do about that. If RealTemp is taking 30 seconds to start showing you temperature data then it sounds like there is something wrong with your computer. It always takes less than a second to show temp data on my computer when I start RT.

As I mentioned before, some nanny software might consider reading temperature data directly from a processor to be suspicious activity. If you push deny and RealTemp doesn't read temps anymore then you have two choices. Tell ZoneAlarm to allow RealTemp to go crazy or you can remove RealTemp and try a different solution. I use an older version of ZoneAlarm and love it. Luckily it doesn't have any of the new "Suspicious Behavior" code in it.

Xello: Generally my idle temps are 32,32,36,25 so i wonder if they should actually be 32,32,32,25

Your cores are numbered from 0 to 3. Core2 reading 36 is likely stuck and the other pair sitting nicely at 32 might be stuck as well but they might just need a calibration adjustment. You can't assume anything but it's certainly not unusual for 3 of your 4 cores to be stuck. All of these sensors seem to get stuck at some point. The lucky ones have processors with sensors that stick at a temperature lower than their normal operating temperature. You should do that quick Prime test I mentioned before to see if the data looks suspicious and then try running the Calibration test as outlined in the docs. My best initial guess is that core3 is probably reading too low and will need to be calibrated upward and I wouldn't do anything with core2 if it is stuck.

[WoZZeR999]

Just thought I'd let you know, the folder name in 'RealTemp_2.70.zip' is 'RealTemp_2.60'. This is at TPU.

[unclewebb]

Thanks for noticing that WoZZeR999. The new files are in that folder so I'll ask W1zzard to update the folder name when he gets a chance to prevent any confusion.

[-X-hellfire]

Thanks W1zzard.

-X-hellfire: I just remembered that there is one time when RealTemp will try to connect to the internet. Your warning would only make sense if you opened up the About... box in RealTemp and clicked on the TechPowerUp logo. That would then try to open up your default browser, Firefox, and open up the RealTemp home page at TechPowerUp. That's the only time RealTemp tries to connect to the internet and it doesn't send your latest Visa statement or anything else to TechPowerUp. It should just open up that page and that's it.

Here's the C++ code:

// Close the About window and head to the TechPowerUp website
void CAboutDlg::OnTechbutton() '
{
	ShellExecute( NULL, _T("open"), _T("http://www.techpowerup.com/realtemp/"),NULL, NULL, SW_SHOWNORMAL );
	SendMessage(WM_CLOSE,0,0);
}

I'm not sure about using SendMessage to close the About dialog after someone clicks on it but it seems to work so I left it. If anyone knows a better way to accomplish the same thing then let me know. Maybe I'll eliminate the automatic closing of this window so I won't have to send any of these suspicious messages.

Now if you have a virus on your computer and it is infecting your .exe files like RealTemp then there isn't much I can do about that. If RealTemp is taking 30 seconds to start showing you temperature data then it sounds like there is something wrong with your computer. It always takes less than a second to show temp data on my computer when I start RT.

As I mentioned before, some nanny software might consider reading temperature data directly from a processor to be suspicious activity. If you push deny and RealTemp doesn't read temps anymore then you have two choices. Tell ZoneAlarm to allow RealTemp to go crazy or you can remove RealTemp and try a different solution. I use an older version of ZoneAlarm and love it. Luckily it doesn't have any of the new "Suspicious Behavior" code in it.



Your cores are numbered from 0 to 3. Core2 reading 36 is likely stuck and the other pair sitting nicely at 32 might be stuck as well but they might just need a calibration adjustment. You can't assume anything but it's certainly not unusual for 3 of your 4 cores to be stuck. All of these sensors seem to get stuck at some point. The lucky ones have processors with sensors that stick at a temperature lower than their normal operating temperature. You should do that quick Prime test I mentioned before to see if the data looks suspicious and then try running the Calibration test as outlined in the docs. My best initial guess is that core3 is probably reading too low and will need to be calibrated upward and I wouldn't do anything with core2 if it is stuck.

I have never opened the "about" in RT dint know it had any, they often show when you right click on the taskbar icon


must be something else that triggers this strange RT behavior on my computer...

btw my firewall or adsl modem is probably near death as I have to wait about 15-20 minutes when they are switch on before even trying to start the computer, as I can see on the LEDs on the adsl modem that it has problems connecting, sometimes rebooting it several times, as all LEDs go blank for a sec, then it stabilize and another on/off of it and turn on the computer it will work dl/stream video/audio all night at up to the max limit of about 500kb/s without any hickups or anything...


think that PG2 act funny due to the adsl modem needs some time to heatup or something when it´s been off

I have put a dedicated 120m fan that cools them down as its in the middle of a heatwave in sweden, 33C sometimes, to keep them from overheating


I read somewhere that some usb-devises could make the computer behave strangely...

The T-Balancer(s) , with its two applications, has some usb problems as once in a while I can hear the usb-plugin/unplug sound and also it crashes while running or when the computer is shutdown

Antoher usb-devise is a 1-2 usb hub which are connected to two usb printers, the laser printer, HP 6P, got an usb-to-Parallel Link
http://www.sandberg.it/product.aspx?id=133-00

but its strange that the 1-2 hub gets "off-line" from time to time, hav to reboot computer to get it "on-line" again, very annoying while printing out stuff

all im saiying is that it the usb problems might cause some weird windows xp behaviors

[WoZZeR999]

Thanks for noticing that WoZZeR999. The new files are in that folder so I'll ask W1zzard to update the folder name when he gets a chance to prevent any confusion.

Yeah, I had noticed that it had the correct files. I did get confused for a second, and thought it was a cached file, but when I saw the file name and build date I figured it was just the folder name.

[WoZZeR999]

I have my T-Bal connected directly into the USB header, and have not noticed it disconnecting/reconnecting. You may want to try a re-install, and if you are still having problems, find out if it's a hardware related issue.

[-X-hellfire]

I have my T-Bal connected directly into the USB header, and have not noticed it disconnecting/reconnecting. You may want to try a re-install, and if you are still having problems, find out if it's a hardware related issue.

im using two internal usb-headers...
T-Balancer XL ...in the roof of the case
T-Balancer XL ...in the bottom of the case

the one in the roof has two internal usb-cables connected together to be able to reach the usb-headers on the motherboard, im pretty sure this is causing that the it get disconnected/connected, hav also prblem conencting more than 4 digital tempsensors as that bus then starts to send wrong temp data, even though they all indicate they are working with each LEDS they are not able to send the digitaltemperaturedata correctly


haven't found any longer internal usb-cable, 1m (3 feet) or more which would eliminate that possible error source

[WoZZeR999]

Are you using the actual USB header with the pins, or a standard usb cable? Since the usb header just uses pins, you could cut each of the 4 wires and extend them with wire connecting them. Use solder and they will not come apart.

[emoners]

@uncle: using the 2.70...seems that if i minimize the window, it stays in the taskbar, unlike previous betas, only the tray temps remain

EDIT: dumb me...there's an option in settings to keep it in the task bar or not. dope!

[Xello]

Well good news, the glitch has ceased. Bad news is i don't know what fixed it, the new version or the driver update

I tried your test webb, this is me at idle:

17:27:52 32 32 36 25
17:27:53 32 32 36 25
17:27:54 32 32 36 25
17:27:55 32 32 36 25
17:27:56 32 32 36 25

And here is when i start prime mall FTTs:

17:28:00 32 32 36 25
17:28:01 32 32 36 25
17:28:02 46 46 45 39
17:28:03 50 49 46 41
17:28:04 50 50 47 42
17:28:05 51 50 50 43
17:28:06 51 49 47 43
17:28:07 51 50 50 42
17:28:08 51 50 50 44
17:28:09 51 50 50 44
17:28:10 51 50 49 43
17:28:11 51 51 50 44
17:28:12 51 51 49 44
17:28:13 51 51 50 44
17:28:14 51 50 50 43
17:28:15 51 51 50 44
17:28:16 51 51 49 44
17:28:17 51 51 50 43
17:28:18 51 51 49 44
17:28:19 51 51 50 44
17:28:20 51 51 50 44
17:28:21 51 51 50 44
17:28:22 51 51 50 44
17:28:23 51 51 50 45
17:28:24 51 51 50 45

Mostly stays like that until i stop it:

17:31:05 53 52 52 46
17:31:06 53 52 53 46
17:31:07 53 52 52 46
17:31:08 53 52 52 46
17:31:09 53 52 52 46
17:31:10 53 53 52 46
17:31:11 37 35 36 27
17:31:12 36 33 36 25
17:31:13 32 32 36 25
17:31:14 32 32 36 25
17:31:16 32 32 36 25
17:31:17 32 32 36 25
17:31:18 32 32 36 25
17:31:19 32 32 36 25
17:31:20 32 32 36 25
17:31:21 32 32 36 25
17:31:22 33 33 36 25
17:31:23 32 33 36 25
17:31:24 32 32 36 25
17:31:25 32 32 36 25

I'm not sure what to make of it

[unclewebb]

@uncle: using the 2.70...seems that if i minimize the window, it stays in the taskbar, unlike previous betas, only the tray temps remain

Holy crap, my heart almost fell out of my chest! I'm glad you figured it out. I did a few last minute changes to how RT minimizes and I had this fear that maybe I screwed something up.

Xello: It must have been the driver!

Here's what I see:

17:28:01 32 32 36 25
17:28:02 46 46 45 39
17:28:03 50 49 46 41

The difference in temps between :01 and :02 is
14 14 9 14

The reason core2 only moved 9 degrees while the other ones all moved 14 is because it is stuck at 36.

Comparing :03 to :01 you get
18 17 10 16

Once again core2 really sticks out as being stuck when this test initially started.

This is good to know. Before I can try and come up with some calibration factors you need to drop your MHz down to about 6x333 or 6x266 and drop your core voltage as close to 1.10 volts as possible. Turn up your fans, open your case and try to force those temp numbers lower. You're trying to prove if any other sensors are getting stuck. Even 1C less will prove that core0 and core1 are not stuck at 32C.

[Xello]

This is good to know. Before I can try and come up with some calibration factors you need to drop your MHz down to about 6x333 or 6x266 and drop your core voltage as close to 1.10 volts as possible. Turn up your fans, open your case and try to force those temp numbers lower. You're trying to prove if any other sensors are getting stuck. Even 1C less will prove that core0 and core1 are not stuck at 32C.

I went to 6 x 333 @ 1.1v, my temps are still 32, 32, 36, 25!

[unclewebb]

Did you open your case during this last test? What is your room temperature near your computer and are you air or water cooled?

If stand by mode works you can put your computer into stand by for half an hour or overnight and after you resume you can check RealTemp to see what it reports for the Minimum temperature. When a computer first starts up it is usually a degree or two cooler compared to when it is just sitting idle.

The other 3 sensors don't seem to be stuck but I always like to try and prove it just to make sure. Your temps returned to 32 32 36 25 awfully quickly after your Prime run so I'm surprised that if left sitting at idle that it can't go 1C cooler. Does your neighbor have a cool basement you can go do some testing in?

[Xello]

I'm afraid not! And i'm not sure what my ambient room temperature is, but it's quite cool today here in scotland - my prime run today had 53c max, compared to 58c yesterday. The case was open, but the whole thing is water cooled anyway, cpu and northbridge included (X48 blackops motherboard). The northbridge temperature did drop to 35c idle when i switched to 6x333 1.1v cpu, it usually idles at 39/40. :edit: My northbridge is currently showing 39c during full load prime test though, it has peaked 51c this past week with the hot weather so i think it's probably just down to it being a cool day.

[unclewebb]

If you're water cooled then we need to compare RealTemp to your water temp. That helps explain why your core temps drop so quickly back to normal. Do you know how accurate your temperature reading for your northbridge is? If the northbridge temperature dropped but the cores didn't drop then it makes you wonder if the core sensors are stuck.