Real Temp - New temp program for Intel Core processors

[rge]

Maybe users will start comparing Min and Max VID now.

I think you may be right.

I have pics of my E8400 with VID 1.1125 when it first came out, but that was older version of coretemp, which was probably in error.

And I can make it now 1.15 to 1.225.

I wonder if it is possible to have software report VID range, min to max?

[unclewebb]

I wonder if it is possible to have software report VID range, min to max?

I'll get right on that! I don't know if it's possible to read Min VID from the processor unless your motherboard and the settings you're using supports that. At least I know what to look for now when reading some registers.

loonym: If RealTemp is reporting Minimum VID then if you start up Prime or whatever it should jump and report the Maximum VID. The T7200 mobile chip I tried showed 4 different VID values depending on load when transitioning between Minimum and Maximum VID.

[loonym]

Interesting. On this board and the x3360 it doesn't seem to change depending on load and stays constant at 1.15 according to realtemp or any other software, everest (1.1), coretemp (1.0375), or cpuz (also 1.15)

[unclewebb]

I use RMClock 2.30 to toggle C1E within Windows. The newer version of this program isn't stable on my computer. Maybe on your board, the C1E setting might make a difference.

[jtaber79]

Hey guys,

New to the forums here. I've been reading through this entire thread for the last couple days and found that it made for some very good reading/info.

It's been a while since I've been big into overclocking ... I had the same PC for about 5 years -- and recently found a little bit of money to upgrade. It's not a powerhouse PC, but it works for what I use it for (mostly development and surfing).

Unfortunately, I've have been blessed with an E6400 with L2 stepping. CPU-z reports it as being a Conroe core (which I somewhat believe), however it only has 2MB of L2 Cache (which also leads me to believe it may be an Allendale). In any case, I'm having a difficult time determining what the correct TjMax for this chip is.

Coretemp 0.99 gave me idle temps of 45C/46C and load temps of 60C/60C (that was with a 100C TjMax I believe). I came across this site and found RealTemp and found that it uses 85C. Following the calibration instructions, RealTemp is now showing me idle temps of 31C/30C and load temps of 46C/46C. That seems about right as the ambient temp in my office is 24C.

Before I got my E6400 I had a P4 531 (Prescott) which I'm inclined to believe it ran much hotter than the E6400 and and my Gigabyte G-Power Pro kept it running at 39C idle and around 54C under load. I'm currently using AS5

Since you guys seem to know your stuff, so I'll ask. Do my temps seem right?

Screenshots:

[unclewebb]

RealTemp 2.64 is available from the beta testing site.
http://www.fileden.com/files/2008/3/...alTempBeta.zip

It includes a new mini mode which may not be as good looking as a Vista gadget but it's useful all the same. Just position RealTemp where you want it and then double click anywhere on the dialog to enter mini mode and double click again to go back to normal. It also works with the Always on Top option as long as you set that first.

The Windows Task Manager also dumps the title bar, etc. after a double mouse click so I shouldn't get in too much trouble for creating a hidden feature.



I'll try to add an option so you can start up in mini mode for the next release.

jtaber79: Go into the cpuz.ini file and set:
Sensor=ON
so it reports your actual core voltage and not VID. I believe that RealTemp is closest to the truth for the L2 series but I'm kind of biased and all of the competition seems to disagree. Oh well. If you've gone through my Calibration procedure as outlined in the docs then you'll probably come to the same conclusion that there's no chance that TjMax=100C for your processor. CoreTemp started this nonsense when the E4300 L2 came out to try and correct for sensor issues at low temperatures but this was a mistake. Everyone else blindly followed along.

[WoZZeR999]

Looks great! In mini mode you should be able to move the window by clicking anywhere on the box. Don't know if it's a pain to implement that, but it could be helpful.

[roflcopter]

Great little piece of software. I only have one small problem: RealTemp (any version, even the newest one) always starts maximized if I put a shortcut in my Start Up folder. Even if I have ticked the Start Minimzed setting or if I start the shortcut with the Minimzed setting under the shortcut properties.

[unclewebb]

WoZZeR999: Good idea. I'll see what I can do.

roflcopter: Adjusting the shortcut properties shouldn't make a difference. I'm using XP and have RealTemp in my Start Up folder and it starts up Minimized in the System tray area. Where do you have the RealTemp folder located? Some users have problems because they are using a limited account and then stick the RealTemp folder in a location like C:\RealTemp that they don't have the priviledge to read or write to so their settings are not saved. Try putting the RealTemp folder on your Desktop and make sure Start Minimized is checked. Give me some more details if this doesn't work for you like what OS you're using so I can investigate further.

[ferdz_33]

@unc,

how come after downloading 2.64 and overwriting the old files (2.63). i still get 2.63??

[unclewebb]

how come after downloading 2.64 and overwriting the old files (2.63). i still get 2.63??

This is usually a problem related to having the old version sitting in your cache so the new one doesn't get downloaded. If you are using Firefox and go into the Tools -> Options -> Advanced -> Network tab and click on the Clear Now button then it should work.

[msgclb]

@unc,

how come after downloading 2.64 and overwriting the old files (2.63). i still get 2.63??

Are you using Firefox? unclewebb gave me the following advice to solve it...

http://www.xtremesystems.org/forums/...postcount=1186

I'm wonder if Firefox 3 will have the same problem.

[roflcopter]

roflcopter: Adjusting the shortcut properties shouldn't make a difference. I'm using XP and have RealTemp in my Start Up folder and it starts up Minimized in the System tray area. Where do you have the RealTemp folder located? Some users have problems because they are using a limited account and then stick the RealTemp folder in a location like C:\RealTemp that they don't have the priviledge to read or write to so their settings are not saved. Try putting the RealTemp folder on your Desktop and make sure Start Minimized is checked. Give me some more details if this doesn't work for you like what OS you're using so I can investigate further.

I'm an administrator on my pc so it shouldn't be a user rights problem. I'm using Vista 64. I tried it by placing a shortcut in the startup folder and by adding RealTemp as an startup item in the registry. But this doesn't help and RealTemp always starts maximized. If I check the .ini file startminimized is set to 1.

[WoZZeR999]

Just had another thought (sorry, I just keep coming up with things that I think would be cool to have), when in 'mini mode', have a docking option for corners/edges of the screen (I would say other windows, but I don't want you to be driven insane).

[unclewebb]

roflcopter: I think this might be a Vista 64 compatibility issue. In XP it works fine and I think it will start minimized to the System Tray or Task Bar in Vista 32 OK. Hopefully someone can confirm that. I'll look into doing this another way. Does anyone know if this feature ever worked in RealTemp when using Vista 64? If so what version of RealTemp?

I made a change a while ago so that it goes directly to the tray but that might be causing a problem in Vista 64 so I'll try to find a work around.

[WoZZeR999]

Just tested 2.64 and start minimized worked as it should on vista x64.

[jtaber79]

jtaber79: Go into the cpuz.ini file and set:
Sensor=ON
so it reports your actual core voltage and not VID. I believe that RealTemp is closest to the truth for the L2 series but I'm kind of biased and all of the competition seems to disagree. Oh well. If you've gone through my Calibration procedure as outlined in the docs then you'll probably come to the same conclusion that there's no chance that TjMax=100C for your processor. CoreTemp started this nonsense when the E4300 L2 came out to try and correct for sensor issues at low temperatures but this was a mistake. Everyone else blindly followed along.

unclewebb:
I checked the cpuz.ini file and Sensor=ON is set.

It, nor any other utility (tried Everest as well) will show anything different than 1.325v.

Must be my crappy Intel mobo.

I would have to agree that RealTemp is the closest to truth. I've seen your test with the E2160 and the IR gun. You definitely know what you're talking about.

[unclewebb]

Here's my best test of an E8400. I haven't seen too many real world tests published by the competition.
http://www.xtremesystems.org/forums/...&postcount=573

Looks like CPU-Z is unable to read the actual core voltage on your motherboard. If it can't find a sensor to read it defaults to displaying the VID which is the voltage that the processor is asking the motherboard for but may not have anything to do with the core voltage that it is actually getting. Maybe the author of CPU-Z might be able to help you. He's done an excellent job at correctly reporting core voltage on most motherboards.

[jtaber79]

Turns out I had an old version of Everest.

Looks like the actual voltage is 1.301v with EIST & C1E off and fluctuates between 1.195v and 1.301v with them on.

Everest says the Sensor Type is: Analog Devices ADT7476 (SMBus 2Eh)

I guess I will e-mail CPU-z and see if they can help.

[unclewebb]

Here's something that's very interesting. It's kind of long but worth reading if you own a Quad.

I've been using the same Q6600 for the last few months for testing purposes. I've probably stared and analyzed the DTS data coming out of these sensors more than anyone in the world has analyzed DTS data. Here is a post from yesterday that shows the typical DTS data from this processor.

http://www.xtremesystems.org/forums/...postcount=1444

The DTS data from sensor 0 and 1 has always been exactly the same at idle. Core3 is usually the same as the first two cores or reads 1 higher. Core2 has always been out to lunch and usually reads 4 higher than the first two cores. This has been consistent for months.

Tonight I have a look at my temps and they're all over the place. I turned off the calibration feature and started up CoreTemp and set it to report the raw data coming from the digital thermal sensors so I'd have something to compare to. RealTemp reports this data as Distance to Tj Max.



CoreTemp and RealTemp are both reading the exact same DTS values from the processor but now the sensor on core2 is correct and core1 is wrong. The readings coming from these two cores appears to have swapped.

This makes no sense so I set SpeedFan to display DTS and it reports the exact same thing.



To try to understand what was going on I started up the multi core version of Prime95. By going into the Task Manager you can use Set Affinity and limit what cores it runs on.

Just to clarify things, a Quad core is basically two separate Dual cores. What I found in previous testing was that if I put all of the load on core0 then core1 would heat up to a similar temperature because it's directly connected to core0 while the other two idle cores that were separate would run cooler.

Now when I run this test, all temperature monitoring software shows that when I put a Prime load to core0, its partner, core1 stays cool and core2 heats up which implies that core0 and core2 are now partners. As I move the load to each individual core it's obvious that the readings being displayed as core0,core2 are coming from one dual core and the readings from core1,core3 are coming from the other dual core.

Anyone with a Quad has seen this before where two cores run at one temperature while the other two cores run at a slightly different temperature. There has never been any consistency of what cores will run at the same temperature but now its starting to make some more sense. There is either a bug within Intel Quad processors that when software asks it to read the sensor, it sometimes returns temperature data from the wrong core or there is a bug within Visual C++.

Given that 3 different programs are all reporting the same, the bug looks more like an Intel issue. One more reason not to remount your heatsink 101 times when the temp data is looking a little odd on a Quad. The temp data is odd!

Thank God for RealTemp. If this new pattern stays the same I can just quickly swap calibration factors for core1 and core2 and I should be good for the next few months. The competition continues to ignore the numerous problems with getting accurate data out of these sensors, and not just the 45nm sensors.

[WoZZeR999]

I had noticed that alot. Either Core 0 and Core 1 were linked, or Core 0 and Core 2 were linked. I have yet to see Core 0 and Core 4 linked.

[unclewebb]

WoZZeR999: You don't have to look at too many screen shots to see the linking of the cores in a Quad. I always thought that Quads came in two varieties with either core0/core1 linked or core0/core2 linked. It's easy enough to test for this by moving Prime, using the Task Manager, to each core as it is running. Now I've got one processor that after 3 months of use has suddenly changed.

Here's a screen shot from the end of March when I first installed this Q6600 showing the original behavior.
http://www.xtremesystems.org/forums/...&postcount=567

Now core1 and core2 have swapped and all temp software is reporting the same so it wasn't a change that I made to RealTemp. Maybe Quads are designed to do this after so many hours of service. I've seen a lot of hard to explain stuff coming from these sensors but this latest issue is a mystery.

[unclewebb]

RealTemp 2.65 is available in the beta section:
http://www.fileden.com/files/2008/3/...alTempBeta.zip

Just a quick update so that RealTemp can report current VID as well as Maximum VID by clicking on the toggle switch in the upper right. If you go into the Control Panel -> Power Options and set your processor to Portable/Laptop, you might see a difference in these two values.

Here's what my board shows with two instances of RealTemp running:

[loonym]

RealTemp 2.65 is available in the beta section:
http://www.fileden.com/files/2008/3/...alTempBeta.zip

Just a quick update so that RealTemp can report current VID as well as Maximum VID by clicking on the toggle switch in the upper right.

Outstanding!!! Another wonderful update to an already great release.

[ferdz_33]

@unc,

thanks for answering my question above. yes, i am using firefox 3 but before i posted that i did clear my cache but still no go. so what i just did is use a download manager (FlashGet) and it worked fine. also tried IE and it downloaded the latest one 2.65.

again keep up the good work and nice to see you also added the multipliers. cheers!