Real Temp - New temp program for Intel Core processors

[balewolf]

theirs my 3700 run i thought my temps were off cause you would thk it would run cooler with the low vcore i seen others with higher vcore run lower with same room temps and heatsink.

[lefy]

am i doing something wrong here ?????

42


lol


no, seriously. i think it's already been addressed, but it would appear that we are victims of our BIOS with the abit IP35 pro. sounds like uguru is the only util worth using for temps/volts/clocks. even my cpu-z results are unreliable with my qx9650... it shows my vcore as 1.64 regardless of actual bios settings

i've got another mobo en route, but i digress to far


... this real temp util is great. thanks again for making it

[unclewebb]

balewolf: Keep in mind that these sensors were not designed or calibrated by Intel to provide accurate idle temperatures so get out of the habit of comparing your CPU idle temps to others. The typical margin of error is between +/- 5C to +/- 10C at idle when uncorrected so idle comparisons are meaningless. These sensors can go either way and be reporting too high or too low at idle. The majority of the 65nm sensors like in your Quad seem to report too low but there's no way to pick up a processor and know what it's going to do by looking at it.

The sensors in your chip actually look really good. A Quad consists of two Dual cores and your core0/core1 are well balanced and your core2/core3 are also well balanced. It's rare to find a 45nm Quad that looks anything like that. They're more like 4 random number generators.

Just go into the Settings window and try using 1.0 for core2 and core3 for calibration factors. After this, your temps from idle to full load will be more accurate with RealTemp than any other program can provide you.

[balewolf]

Ya im not hung up on idle temps my main thing is most other quads using more vcore with same or lesser heatsinks run cooler thats why i was thinking my sensors were off mine has been lapped flat and mirror finish so has the tuniq tower. And today and tonight playing around with differt fsbs my temps seemed to change they got lower when priming compared to what they were showing at same clocks i got a wierd q6600 lol.

[unclewebb]

balewolf: What are you using for thermal paste and how do you apply it? I find a little too much is better than not enough and I like to spread AS5 in a thin rectangular area directly over where the cores lie underneath. Intel has some good pictures of how a Quad looks when the IHS is removed.

I find this method results in consistent temps from the moment you install the heatsink. Other people talk about how the TIM has to set up and then the temps will come down but I don't see much difference using this method from day 1 to day 101. I've never had great results using an amount equal to the head of a pin or whatever it is.

My Q6600 and old motherboard are not a happy match. I can run my E8400 for a suicide Super PI run at 4500 MHz but so far I haven't had my G0 Quad much above the low 3000 MHz range. I haven't really pushed it but I know core3 is always eager to fail Prime. At your current voltage of about 1.30 volts why don't you drop your MHz down to default of about 2400 MHz and run Prime95 small FFTs on all 4 cores. I'm curious to see how our Quads compare. We're both using the same heatsink and I know exactly what temperature my Quad runs at. Include your room temperature and whether your case is open or closed and use the +1 calibration factors like I suggested on core2 and core3 to help get your idle temps equalized. These calibration factors will also improve the accuracy of your full load temps as well. Only about 1C difference at full load but you seem interested in accurate temps so you might as well use it.

Some processors do run hotter than others when everything is equal. If the IHS and heatsink are both square in both situations then it might be a problem with how good the IHS to core heat transfer is.

Post a screen shot of your CPU at idle and one after about 5 minutes of running Prime and I'll do the same. Time to vacuum out the dust bunnies so I can show you some honest temp numbers that you can compare to.

Edit: I didn't feel like rebooting and playing with my settings so here's what I get for my recent 24/7 settings at a room temperature of 22C.





The Minimum temperature was recorded this morning when I first started up from Stand By mode and the Maximum was recorded before I got up and turned my CPU fan up to high. After the fan speed went up, core0/core1 settled in between 52C and 53C.

[PCTwin]

shot of my qx6700 B3 with Hwmon and Realtemp. I dont use hwmon anymore as the motherboard temps are crazy, jumping up and down. The core temps seem to agree with Coretemp as they are always 15c above realtemp.

[unclewebb]

PCTwin: I took some abuse when the first version of RealTemp suggested something crazy like TjMax=85C for the B3 Quad processors but your RealTemp reported numbers look about right.

I'm really disappointed in HWMonitor. I thought because it was related to CPU-Z that it would be in the same league for quality and accuracy but obviously it is far from that. If you were actually feeding your DDR2 4.66 volts it's surprising they all didn't go POP!

For any new users that don't understand the different temps from different programs I'll type out the formula one more time:

Reported Temps = TjMax - Digital Thermal Sensor value

Most programs do a decent job of reading data from the Intel DTS sensors but if one program assumes TjMax=100C and the other program assumes TjMax=85C then the reported temps will be different by exactly 15C.

[PCTwin]

decided to clean out my box and try calibrating. you've probably got all this info but here's my results. my thermometer shows 22c room temp.

[unclewebb]

PCTwin: Thanks for showing everyone how easy it is to get your idle temps balanced. A lot of people don't understand that these calibrations that are done at idle also improve the accuracy of your load temperatures up to about 60C. I'm going to try running my Q6600 at 1600 MHz and 1.10 volts and see how that compares to 1600 MHz and 1.30 volts.

Edit: Almost surprisingly, there is very little difference in idle temps even after dropping the core voltage significantly. With a room temperature of about 26C, case open, Tuniq on high, I got an average idle temperature of 33.5C at 1.304 volts.



Dropping the core voltage down to 1.080 volts only dropped the average idle temperature down 1C to 32.5C. That's exactly what you're seeing balewolf so I guess your processor is pretty normal.

[balewolf]

Ok i did what you asked but i waset running same fsb and i need alot less vcore but my temps arent that much lower then yours really. I got a nice low vid quad that will do 3700 on 1.31 vcore but temps been holding me back pushing it much further folding Ambs are 18-22c no higher.

[unclewebb]

balewolf: I think you put the calibration factors on the wrong cores. Here's what I recommended.

"Include your room temperature and whether your case is open or closed and use the +1 calibration factors like I suggested on core2 and core3 to help get your idle temps equalized."

Room temperature directly effects core temperatures so when you're doing testing you can't say things like it was somewhere around 18C to 22C. That's like telling the cop when he pulls you over, "What's the problem officer, I was only doing somewhere around 70 mph to 100 mph."

Show a screen shot of your RealTemp Settings window. Maybe there's a problem with my program. Post #1611 just above by PCTwin is an excellent example of how to equalize idle temps across all 4 cores.

[balewolf]

I havet calibrated it all yet thats how it is default and on the room temp it varies im in the back of the house and the temps go up and down ac is set on 19c and at night my room probly is 19c but daytime its closer to 22c. Its not just your program that shows a large temp jump every other one does to. I thk this quad just is a hot one my old one idled and ran cooler @ 3400 with 1.41 vcore it idled at 25c across all cores and loaded no higher then 56c priming.

[Shadowthor]

You have the best VID for the Q6600. So those are the idle temps?

[balewolf]

Those are load temps running folding 450x8 3600mhz.

[-X-hellfire]

when will RealTemp support/show temp at the clock?

in the way other programs do... Everest, Coretemp etc

...

[msgclb]

when will RealTemp support/show temp at the clock?

in the way other programs do... Everest, Coretemp etc

...

Do you mean like this?



Click the Setting button and look under System Tray Settings.

[unclewebb]

when will RealTemp support/show temp at the clock? in the way other programs do... Everest, Coretemp etc..

Fortunately, RealTemp will never show completely inaccurate temperatures in the System Tray area like Everest and CoreTemp do.

[Volkswagen]

so is this the program to use for accurate temps?

[R@ptor =]]]

I don't know what program to use yet...
core temp or real temp?

[-X-hellfire]

Do you mean like this?



Click the Setting button and look under System Tray Settings.

Yeah, great unclewebb have fixed that while I wasn´t looking, I´m been following this thread forever and ever it seems I must have missed it

Fortunately, RealTemp will never show completely inaccurate temperatures in the System Tray area like Everest and CoreTemp do.

Thanks for your tireless effort of debunking the blue box of semi-random temp-numbers


I will dl & install it ASP

[unclewebb]

It's summer time here, the kids are out of school and I've been too lazy to finalize the next release. The latest beta is pretty close to finalized.

http://www.fileden.com/files/2008/3/...alTempBeta.zip

Download, unzip and copy the new RealTemp.exe into your present RealTemp folder and overwrite the old RealTemp.exe with the new version.

R@ptor =]]: If anyone isn't sure what temp program to use then at least start here:

http://www.xtremesystems.org/forums/...postcount=1600

and do some research and then ask the competition to provide you with some of their research.
Don't hold your breath waiting for a reply.

[Shadowthor]

Trying out the beta, seems good so far .

[msgclb]

It's summer time here, the kids are out of school and I've been too lazy to finalize the next release. The latest beta is pretty close to finalized.
...
Download, unzip and copy the new RealTemp.exe into your present RealTemp folder and overwrite the old RealTemp.exe with the new version.

Did you use the same version 2.68? I've cleared my cache and I believe that I'm downloading the version released on the 26th of June. I even downloaded it with IE 7 and got the same version.

[loonym]

Did you use the same version 2.68? I've cleared my cache and I believe that I'm downloading the version released on the 26th of June. I even downloaded it with IE 7 and got the same version.

Same here. I was expecting to see a 269

[unclewebb]

Sorry for the confusion. Version 2.68 is still the latest beta. Production slows down during the summer! I just re-mentioned this beta since it sounded like some users were well behind in the version they were using.