Real Temp - New temp program for Intel Core processors

[JohnZS]

You certainly have been busy Unclewebb
I best see if I can have a look at your latest builds, I'm still happy with the 2.75 build.
Keep up the good work and hopefully (fingers crossed) intel release USEFUL information this time instead of marketing their upcoming Core i7
John

[unclewebb]

My original theory when I first started out was that 45nm and 65nm sensors were very similar with similar amounts of error. Intel's release of TjMax for 45nm showed me that I completely underestimated the amount of error in the 45nm sensors.

I couldn't imagine that Intel would be using a sensor that could be off by as much as +/- 10C at 50C but that is their official spec for the 45nm sensors used on the Intel Atom CPUs. Based on the data I've seen and user feedback, that amount of error seems to be what the 45nm Desktop E8000 and Q9000 sensors also have. Some sort of calibration is not an option for 45nm.

With the correct TjMax for 65nm, users should have very accurate temperatures from idle to TjMax. The 65nm sensors are far more accurate and rarely if ever suffer from saturation or sticking in the temperature range that most users operate at. The amount of slope error in these sensors is probably one quarter of the amount of error in the 45nm sensors. I think the latest beta of RealTemp is using the correct TjMax for most 65nm processors but it will be good to finally get this information from the manufacturer so all temp programs can finally agree at least to one part of the equation.

There are still a few minor things to do to RealTemp but it works OK so I haven't been too motivated lately. I've been having too much fun playing with my new Reapers! Not too bad for the old P5B.

[burebista]

[...] so I haven't been too motivated lately.

Grrr...Hmmm...Pffff...
Nice work man. Both for RT and your OC.

[ZL1Killa]

not to hijack or anything, on my recent laptop purchase, g50v-A2 ASUS laptop, I was playing crysis warhead and with this program it told me my temps were 65 C and 67 C on my CPU... is this hot for intel? I know its hot for AMD...and its hot in general, freaking 150 degrees F.

[unclewebb]

... is this hot for intel?

All I know is that it's too hot for my lap!

Core based CPUs are designed to operate very reliably at some rather extreme temperatures. If Intel thought your CPU was too hot they'd lower the thermal throttling and thermal shut down points so that your CPU would never get too hot.

On a 65nm laptop, thermal throttling doesn't start to happen until almost 100C and for 45nm mobile CPUs, it is 5C hotter than that. Thermal shutdown doesn't happen until approximately 125C. If Intel thought your CPU was going to turn into a puddle of silicon then they'd obviously lower those temperatures.

As long as your computer is stable and running below the thermal throttling point, core temps are unimportant.

[ZL1Killa]

All I know is that it's too hot for my lap!

Core based CPUs are designed to operate very reliably at some rather extreme temperatures. If Intel thought your CPU was too hot they'd lower the thermal throttling and thermal shut down points so that your CPU would never get too hot.

On a 65nm laptop, thermal throttling doesn't start to happen until almost 100C and for 45nm mobile CPUs, it is 5C hotter than that. Thermal shutdown doesn't happen until approximately 125C. If Intel thought your CPU was going to turn into a puddle of silicon then they'd obviously lower those temperatures.

As long as your computer is stable and running below the thermal throttling point, core temps are unimportant.

they sure have raised the operating temperature of things in general lately...at least to my knowledge. that is freakin hot, lol.

But yeh, the laptop is running fine. I usually have it sitting on a desk or my large wooden cuttin board with my laptop cooler in bed, so it never really sees my lap unless i'm traveling.

[MacClipper]

Just a small update that took me far too long to get working more or less right in both XP and Vista. I hope there's a happy user somewhere that will enjoy it.

Yup, simply lovely so thanks again.

[DonInKansas]

Thanks Don. I've been looking into adding a touch of Aero to RealTemp for the Vista Ultimate guys. I promise not to get too carried away if I go for it.

I found an interesting freeware program called ClearTweak that lets you sharpen up your fonts. It's simple to use and the price is right.

I run Ultimate64 if you need a guinea pig.

[traveler]

Version 2.70 - Released July 31, 2008



[snip]

I notice that the core temps for all 4 cores are all within a couple of degrees of each other.

Is that the way it should normally be? At idle? At load?



__________________
Asus X48 Rampage Formula / Intel Q9450 / 2 x 2GB OCZ Reaper HPC PC2 8500 / VisionTek Radeon HD3870X2 OverClocked Edition /
Western Digital Raptor X Hard Drive / Two Western Digital Caviar RE2 WD5001ABYS 500GB 7200 RPM SATA 3.0Gb/s Hard Drives /
Creative SoundBlaster X-Fi Elite Pro 7.1 / Creative Gigaworks S750 7.1 Speaker System /
Lian Li 343B cube case / PC Power & Cooling Silencer 750 Crossfire Edition

[unclewebb]

I notice that the core temps for all 4 cores are all within a couple of degrees of each other.

Is that the way it should normally be? At idle? At load?

If your 45nm Quad looks nothing like the screen shot of my Q6600 that you posted then don't be surprised. There is far more error in the 45nm sensors compared to the original 65nm sensors. On a 45nm Quad, reported temperatures are typically all over the place at idle and full load. At idle you also get problems with sticking sensors so one core stuck and reading 10C or more higher than the others is also pretty typical.

These sensors weren't designed to accurately report core temperatures. Intel designed them to trigger thermal throttling and thermal shut down. There is a small amount of error where Intel calibrates them near TjMax and they can be off by 10C or more at idle even when they aren't sticking.

On a 45nm Quad when some sensors are reading too high and other ones are reading too low and some are getting stuck, you can end up with what looks more like 4 random numbers.

My calibration method involves reducing the FSB to its default which is 333 MHz for 45nm, lowering the multiplier to 6.0 and reducing the core voltage to approximately 1.10 volts. This will reduce the heat output of a CPU to a minimum. At idle, with your CPU Usage mostly at 0% or 1%, it's my opinion that the actual temperature of your 4 cores should be pretty much equal. On a CPU with properly functioning 65nm sensors this is what you typically see.

I use Prime95 small FFTs to test at full load. Intel Burn Test / LinPack will produce more heat but I've found that it is not as consistent as Prime95. When running Prime95 on a Quad, it is typical to see the first two cores at one temperature and the second two cores at a slightly different temperature. This might be a small amount of sensor error or more likely just how the IHS is making contact with the cores.

RealTemp 2.79.8 is available here:
http://www.fileden.com/files/2008/3/...alTempBeta.zip

You can read the documentation here:
http://www.techpowerup.com/realtemp/docs.php

It's overdue for an update but the part about calibration should still be OK.

If you take the time to calibrate RealTemp to your 45nm Quad then at idle and at full load your reported temps should be a lot more consistent and look more like what the 65nm Quads look like. If you have one or more sticking sensors then you won't be able to calibrate those cores to accurately report idle temperatures. If you need some help calibrating then just send me a PM.

[randomizer]

Hmm... I'll need about -20C of correction on both cores or a new CPU.

[Demo]

Just a quick check, 100c TJmax for a Q6600 G0 right ?

[WaterFlex]

Anybody knows, what is Xeon E3110 TJmax?

[unclewebb]

randomizer: Intel discourages trusting the 45nm sensors at any temperature below 50ºC because they tend to become "saturated" which is a polite way to say they get stuck. Looks like your sensors are living up to their expectations! At least at full load they seem to be working OK. You just need to run Prime95 small FFTs 24/7 to keep your core temps up into a range where they can be measured more or less accurately.

Demo: If you click on the Settings - Defaults button in RealTemp, it should be using TjMax=100ºC for your Q6600 now. The rumor is that Intel plans to officially release TjMax for the 65nm processors this month.

WaterFlex: Intel didn't officially release any information for the 45nm Xeon chips at their August IDF so a person has to work backwards and take a good guess for the Xeon. The original E8400 - C0 is officially TjMax=100ºC. It has a thermal spec of 72.4ºC which is the same thermal spec for both the C0 and E0 E3110 CPUs. My best guess would be that your E3110 is also TjMax=100ºC.

A lot of users think that using the correct TjMax is automatically going to give them accurate temperatures but that idea is wrong. There is so much error in most 45nm sensors, both at idle and even at the calibration point near TjMax, that you need to do some sort of user calibration to have any hope of reasonably accurate temperatures and unfortunately, that's the best you can hope for.

Intel hasn't released enough information about these sensors for any software developer to translate the data coming from the on chip sensors into 100% accurate core temperatures. The 45nm sensors simply aren't good enough for that. Having said that, I still think that a RealTemp type calibration is better than nothing and will result in some reasonably accurate reported core temperatures.

[WoodButcher]

Having said that, I still think that a RealTemp calibration is better than anything else and will result in some reasonably accurate reported core temperatures.

fixed.

[unclewebb]

Thanks.

[randomizer]

randomizer: Intel discourages trusting the 45nm sensors at any temperature below 50ºC because they tend to become "saturated" which is a polite way to say they get stuck. Looks like your sensors are living up to their expectations! At least at full load they seem to be working OK. You just need to run Prime95 small FFTs 24/7 to keep your core temps up into a range where they can be measured more or less accurately.

Ah, but I am using an E6600 (ie. 65nm).

[sirheck]

Ah, but I am using an E6600 (ie. 65nm).

Time to upgrade ya manegy cat.

[randomizer]

Time to upgrade ya manegy cat.

That would mean I can no longer provide further twists to the story. Can't have that

[unclewebb]

randomizer: I must not have had my eyes open when I looked at your screen shot. Too busy admiring RealTemp! I thought all of the 65nm sensors were damn near perfect.

[randomizer]

They probably are. Both cores are +/-2C of each other when under equal loads which is often not the case with 45nm processors. It is quite possible that my IHS makes poor contact with the cores such that both cores still remain the same temp as each other but higher than they otherwise would have been. I have done the razor blade test and the exposed side of the IHS is not convex/concave to any significant degree (I won't go as far as saying it's flat, because they're never really flat). I've reseated the cooler a thousand times (and my cooler is a real pain to do it with), and used both stock and aftermarket coolers with similar results. The only thing left to try is a proper cleaning off of the TIM. The last time it was properly cleaned was in the fab; I've never put a drop of alcohol on it, I just clean it with a rag until no more TIM rubs off. I wouldn't think I'd improve my temperatures by 20C+ though.

[unclewebb]

If anyone comes across the official TjMax info for 65nm then do us all a favor and post it here. The rumor is that Intel plans to release this info at this week's IDF conference. They're only a couple of years late.

[emoners]

@uncle: the latest version is the 2.79.8 right?

[Dua|ist]

I have another Tjmax question. This might sound weird/stupid but knowing Intel this might be as well true. If the newer E0 stepping is supposed to consume less power (better OC results with lower vcore are a good proof to that) then why so many people report higher temps on their E0 CPUs compared to C0 (including me )?
One russian site compared E8500 C0 & E8600 E0, and according to their measurements, E0 does consume less power at stock (both idle & load, despite a higher speed of E8600). What's more interesting they compared them at same frequencies and voltages (450 x 9.5 = 4275 Mhz @ 1.4v) and once again the E0 E8600 consumed 9W less power under OCCT load. But its max temperature on OCCT graphs was 2C more than that of a E8500. I know all CPUs are different, etc., but (much) lesser power consumption just cannot cause 2C hotter temps, right? Can't it be that Intel once again lowered Tjmax values for E0 CPUs to something like 90-95?
P.S. The article I mentioned is here: http://www.fcenter.ru/online.shtml?a...ssors/24678#04. It's in russian but one can understand the wattage table and graphs.

[carepolice]

I have another Tjmax question. This might sound weird/stupid but knowing Intel this might be as well true. If the newer E0 stepping is supposed to consume less power (better OC results with lower vcore are a good proof to that) then why so many people report higher temps on their E0 CPUs compared to C0 (including me )?
One russian site compared E8500 C0 & E8600 E0, and according to their measurements, E0 does consume less power at stock (both idle & load, despite a higher speed of E8600). What's more interesting they compared them at same frequencies and voltages (450 x 9.5 = 4275 Mhz @ 1.4v) and once again the E0 E8600 consumed 9W less power under OCCT load. But its max temperature on OCCT graphs was 2C more than that of a E8500. I know all CPUs are different, etc., but (much) lesser power consumption just cannot cause 2C hotter temps, right? Can't it be that Intel once again lowered Tjmax values for E0 CPUs to something like 90-95?
P.S. The article I mentioned is here: http://www.fcenter.ru/online.shtml?a...ssors/24678#04. It's in russian but one can understand the wattage table and graphs.

Those E0 And C0 that you compared did they have same VID?