RealTemp General Discussion

[unclewebb]

Here's some testing I did using the new Intel approved TjMax=100C for my E8400. I also did this same test with a Q6600 - G0 with TjMax set to 100C and got the same results.

I played around with it for half an hour and even took it a little beyond TjMax just to make sure that it was thoroughly warmed up.

I tried testing at different temperatures and moved the IR gun around and around but the hottest it would ever show was 5C less than the reported temperature of the hottest core.. This is the gradient that Intel says exists between IHS and the core even when the CPU is 99% idle. The above pic was taken 2 seconds after this one.

Once I put it back together I tried doing the RealTemp calibration test. The room temperature near the open computer was 21C and it was running at 1600 MHz and 1.08 volts. With these settings, at idle, an expected core temperature should be about 5C above the room temperature which would be 26C. This is what I get:

When using the correct TjMax, and no additional calibration, my core temperatures are being reported about 8C too high. This is a perfect example of "slope error." JohnZS has one core that reads too high and one core that reads too low so a combined error, when uncorrected, of 15C on his Quad is hardly surprising.

Intel finally releasing the correct TjMax for 45nm hasn't given the enthusiast community as much information as they thought they were going to get at IDF.

Here's a quote from the press release that was posted at Anandtech:

"Armed with this information, seasoned application developers and amateur coders alike will finally have everything they need to implement the most accurate, real-time core temperature display tool possible."

Well, not quite.

 

[dark2099]

Did Intel release the TjMax for the new E0 E8400 and E8500's when they released all the other 45nm chips or are they the same as the other revisions/steppings?

 

[unclewebb]

Intel did not mention that there was any difference in TjMax with the new E0 steppings.

Here's the IDF presentation:
http://intel.wingateweb.com/US08/published/sessions/TMTS001/SF08_TMTS001_100r.pdf

 

[dark2099]

That link doesn't work or doesn't like me.

 

[unclewebb]

Sorry about that. It originally worked after the IDF.

I found a different link to this document:
http://rapidshare.com/files/141444140/SF08_TMTS001_100r.pdf.html

 

[burebista]

Uncle, I have a suggestion, but before a small intro.
I have an E8400 cooled by a Ninja rev.B fanless+AS5+ TR Bolt-Thru on a P5Q-E in a P182.

Those are my idle temperatures at 26°C room temperature.

I hardly believe that an E8400 at 1560 MHz and 0.944V Vcore have 43°C at idle with 26°C room temperatures.
Then my full-load temperatures with a small OC at Vcore default.

Three different temperatures (RealTemp calibrated, RealTemp/CoreTemp default and SpeedFan with old TJmax) but one thing is the same in all programs: distance to TJMax (it's obvious because all programs reads same value from MSR 0x19C).

So here it cames my suggestion: to get rid off this chaos with calibrating idle temperatures and different TJMax for 65nm or 45nm CPU's (I know that Intel told us at last TJMax for 45nm CPU's but it is still debatable as uncle show us) we can use distance to TJMax as absolute and correct temperature in full-load. OK we'll still have strange temperatures in idle but all that's matter is accurate temperatures in full-load to see how close we are from throttle point.
We can define some ranges (just an example)
- distance to TJMax >55°C. Cold CPU
- distance to TJMax between 25°C and 55°C. Warm CPU
- distance to TJMax between 10°C and 25°C. Hot CPU.
- distance to TJMax <10°C Very hot CPU, approaching to throttle point

All I want is that for those who want to calibrate RealTemp for idle and mess with TJMax leave the program as it is, but for those who want to see if they approaching to throttle point make an option in Settings to show in system tray distance to TJMax and to make it more intuitive color coded it. Something like this:
- green. Cold CPU
- yellow. Warm CPU
- red. Hot CPU
- blinking red/magenta. Very hot CPU.

In this way all became simple. Regardless of your CPU (Quad, dual-core, 65nm, 45nm), your room ambient, your load, your TJMax, if you see bigger numbers you're OK with a cold CPU, if you see small numbers you're in load but still OK and if you see very small numbers you're close to trouble.

Good or bad idea?

 

[Hayder_Master]

anything like real temp for amd

 

[burebista]

Use CoreTemp. You don't need idle calibration and TJMax madness for AMD.

 

[unclewebb]

burebista: With the latest official TjMax release and Intel admitting that the 45nm sensors have some significant slope error issues, being able to see the DTS reading in the system tray is a good idea. I'll throw that on the immediate things to do list and work on colors for it later.

 

[burebista]

I'll throw that on the immediate things to do list and work on colors for it later.

OK man, thanks a lot.

 

[mtosev]

great app. tnx

 

[fishnchips]

amd dual core fx 64 is not supported why is this?

 

[burebista]

Because this?

RealTemp is a utility to monitor core temperatures on all Intel Core based processors.

 

[fishnchips]

Nuts!!!

 

[unclewebb]

amd dual core fx 64 is not supported why is this?

I haven't owned a platform with an AMD CPU since my E6400 arrived and immediately overclocked to 3200 MHz. At the moment, I don't have any easy way to test and develop for AMD.

The AMD sensors don't have the same issues that the Intel 45nm sensors do so a program like Core Temp should be able to give you accurate temperatures for your AMD CPU. I decided to concentrate on the Intel Core CPUs and there have only been 2 or 3 requests in the last 6 months to support AMD.

 

[Kursah]

One thing that would be nice would be actual CPU voltage reading that some other programs use also...dunno if it'd be a pain to add or not, but some people are mistaking VID reading for actual voltage settings when overclocking. I've seen that happen a few times..I'd personally like to see that as one little extra to an already excellent utility.

 

[mlee49]

One thing that would be nice would be actual CPU voltage reading that some other programs use also...dunno if it'd be a pain to add or not, but some people are mistaking VID reading for actual voltage settings when overclocking. I've seen that happen a few times..I'd personally like to see that as one little extra to an already excellent utility.

I'm one of many who made the mistake, thanks Kursah.

//subscribed//

 

[unclewebb]

CPU-z is probably the most well known utility and does an excellent job on most motherboards at reporting CPU voltage so adding that feature to RealTemp probably won't happen in the near future. In any screen shot, CPU-z is the first thing you usually see.

 

[burebista]

Official TJMax for 65 and 45 nm CPU's from IDF Taiwan. In attachment.

 

[unclewebb]

Thanks burebista for posting that. Now we don't have to worry about that document disappearing from the Intel website like after the last IDF.

I've read over the document and noticed that they talk about a value called Target Tj. Users that decide to use Target Tj as TjMax will find that their reported core temperatures might end up a long ways away from reality, especially for processors like my E6400 B2 which has a Target Tj = 70C according to Intel.

 

[burebista]

Now we don't have to worry about that document disappearing from the Intel website like after the last IDF.

Yep, that's was my idea.
Anyway, now I don't understand nothing from TJMax or that brand new Target Tj.
So my advice come together with yours:

It's just a number and for most people that are unable to calibrate due to sticking sensors or simply don't bother, it's not even a very accurate number. Overclock as high as you want and let stability be your guide.

 

[unclewebb]

Good advice burebista.

Here's how my E6400 looks when using this new Target Tj value as TjMax:

The room temperature was about 11C. Using this new Target Tj, my reported core temperature would be down at minus 6C which is a long ways from reality on an air cooled processor.

 

[unclewebb]

RealTemp 2.84

I just finished a new Sensor Test feature in RealTemp.

http://www.fileden.com/files/2008/3/3/1794507/RealTempBeta.zip

It didn't take long for a user to post a screen shot of what happens when a sensor is sticking.

It's obvious that as his CPU cools, core0 gets down to a Distance to TJMax of 55 and that's it. It's stuck at that point and can't go any lower. When the temperature of the CPU is higher than that, it works perfectly fine.

I added a lot of extra code to try to make setting up this test as simple as possible but for best results it will still help if you follow a few directions.

The CPU Cool Down Test uses Prime95 - Small FFTs to create a consistent amount of heat. If you don't already have it installed then click on the Download Prime95 button and it will take you to the website where you can download either a 32 or 64 bit version. Scroll down to Step3 when the Mersenne - Prime95 web page pops up for a download link.

If you have a Dual Core then you could also use Orthos Small FFTs instead. Prime95 has recently been updated to version 25.7 build 3 so I'm recommending that.

You need to let RealTemp start up Prime95. This will give RealTemp full control of Prime95 so it can shut Prime95 down automatically when it reaches the 0.0% level of this test. The CPU Cool Down Test won't start if you already have Prime95 running. You will need to shut it down first before starting this test.

Before clicking on the Sensor Test button, make sure your computer is as idle as possible for best results. Give Vista some time to calm down before starting a test. The first thing it will do is run the previous sensor movement test to give you a quick idea of how your sensors are doing.

To begin the more thorough CPU Cool Down Test click on the Start button and navigate to the directory where you have prime95.exe installed.

When Prime95 starts up, select the Small FFTs option at the top and click on OK to begin the heat up phase. There is plenty of feedback in the info window so you don't have to remember too much of this.

That's about all there is to it to get this test started. Go for a coffee or kick back for about 10 minutes and RealTemp will do the rest. You can Stop or cancel this test at any time if you get bored.

What you're looking for are sensors that show movement at each stage of this test. If there isn't movement then you might have a sticking sensor or two. If both sensors are moving but moving at different rates then that would show "slope error" that Intel has warned us about.

Do some testing and post some screen shots so I can learn some more about these sensors. The new Core i7 sensors seem to perform similar to the previous 65nm Core 2 Duo sensors. They continue to show signs of slope error.

 

[msgclb]

Here's what I got with my E6750.



I have a Asus P6T Deluxe on a FedEx truck that should be delivered by the end of the week. I have an Intel Core i7 920 and a Thermalright LGA1366 Bolt-Thru-Kit on order but I haven't been able to find the memory that I want to buy. Until I do that makes them very expensive paper weights. When I get this system up and running I'll run the sensor test on them.

 

[a_ump]

i found this for TJmax when just googling it for my q6600

http://www.tomshardware.co.uk/intel-dts-specs,news-29460.html