Real Temp - New temp program for Intel Core processors

[RottenMutt]

Please add Dual Socket Support.

[Ovidiu]

My crappy sensors.... E8400 C0 @3600.

[unclewebb]

randomizer: The Intel OEM cooler, when properly installed with an adequate amount of thermal paste and with all 4 push pins fully seated, should have no trouble keeping an E6600 a long ways from thermal throttling. If you have installed your CPU correctly then your recent test screams out to me that your cores and IHS are not making proper contact. Translation: your CPU is defective.

Edit: I just checked the [H]ard database and my best run with the OEM cooler was 3520 MHz at 1.525 volts bios (~1.45v actual). I think it was one of the top ranked CPUs using the Intel heatsink and fan. I guess no one else was crazy enough to run Prime small FFTs for 4 hours out in their garage in the middle of winter with that cooler. When inside I used to run it at 425x8 ~ 3400 MHz 24/7 with a little less voltage to keep the temps down.

Edit2: The above run was in my basement at ~12C. I didn't need the garage until I was trying to get over 3600 MHz.

I realize I was in a cooler environment but from what you've posted, I can only think that there is something physically wrong with your E6600. They have a 3 year warranty so why not send it back to Intel. Maybe they won't have any replacements left so they'll have to ship you a Core i7 and one of their new boards to go with it. An Intel board that overclocks decently. Who would have thunk it?

Ovidiu: Your sensors aren't bad at all. Neither one of them is sticking. If you calibrate them using RealTemp and this post by rge, you'll get some very accurate temperatures out of them. Your data will help me re-examine my slope correction formula.

RottenMutt: I've been thinking about adding Dual Socket support for a while now but I don't have a platform to test on. Do you have some time to do some testing for me? The quickest thing I can think of is to add a switch to the INI file so you could set that to read core 4 to core 7 and when it wasn't set it would read core 0 to core 3. You would need to run two instances of RealTemp. That would be fairly easy to program and would save me the hassle of totally redesigning the GUI for a very small percentage of users. If that works out then maybe some day in the future I'd think about doing an 8 core specific version. If you're serious then PM me and bug me about this project.

[sakis_the_fraud]

here are my temps.



if i got it right, at the sensor movement test the numbers should be close in order to have good calibrated sensors or i misunderstood?

what about my sensors?

and if someone has the time to explain what is TJMax i would be grateful!

[unclewebb]

sakis_the_fraud: Your sensors look great and your temps are well balanced between your two cores like they should be during this test. If you would like to calibrate your sensors then check out rge's post:

http://www.xtremesystems.org/forums/...postcount=2429

TJMax is the maximum core temperature that these CPUs should be run at. The sensors do not output the temperature of the CPU. They only output how far you are away from TJMax. What RealTemp reports as Distance to TJMax it the raw data from these sensors. RealTemp, Core Temp etc., then take this number and try to convert it to an absolute core temperature. You need to accurately know TJMax so you can convert Distance to TJMax into a core temperature.

Unfortunately Intel hasn't shared this information with the user community until just recently and what they have shared recently is still open to interpretation and debate. There are many processors that don't seem to fit their estimation of TJMax.

Most processors when 2 or 3C away from TJMax will start to thermal throttle or slow down to control their heat output. A slow computer is not a good thing so it's best to leave yourself some temperature head room. You'll also have better stability if you can run your computer cooler.

Now head back and read the first 100 pages of this thread so you can get up to speed on this stuff.
Welcome to XS.

[randomizer]

randomizer: The Intel OEM cooler, when properly installed with an adequate amount of thermal paste and with all 4 push pins fully seated, should have no trouble keeping an E6600 a long ways from thermal throttling. If you have installed your CPU correctly then your recent test screams out to me that your cores and IHS are not making proper contact. Translation: your CPU is defective.

I was not satisfied with the lack of clicking I was getting from some of the pushpins so I decided to reseat it, I was strapped for time originally and didn't get a chance. I most certainly used enough TIM since installing the OEM heatsink. Upon removal I noticed that the TIM had not spread uniformly as I had expected given the temps, so I pulled it off and gave the push-pins some PCB-bending force to make sure I got them in. I retested with an ambient temp around 5C lower than yesterday (probably around 22-23C this time). It looks like no Core i7 for me



I now remember why I got rid of this OEM heatsink originally. The constant ramping up and down of the fan just opening a new tab is so irritating.

EDIT: I just compared this to my original test, and considering the other heatsink is a Zalman 9500 clone it did a pretty poor job of cooling the CPU compared to the stock cooler. The real difference was the blissful constant fan speed.

[mine]

Hi Dave




isn`t it nice ..brandnew 9550 E0 @3,2

[Richard Dower]

Here's mine, Q9450 at 413x8=3.3Ghz:
This good or bad?

[JohnZS]

Realtemp keeps getting better and better, keep up the good work Unclewebb
I managed to run the new Sensor Test on my QX9650 Extreme Sensor Edition Processor.
Prime95 was ran using the option for "Maximum Heat and Power Consumption"

Pretty shocking results?
Sensors all seem to be showing the same (or at least reporting) the same degree of movement, yet the values they output are completely random, only Core0 and Core1 show any relation to each other, the rest might as well be spitting out numbers for fun.
Or have I misinterpreted the results?
John

[Dadi_oh]

Here is my Q6600 G0. Under water. Room temp about 22C (guess). Look about right?

[belial]

here is my e7200, looks like sensor on core 0 doesn't work if the load is 12.5% or lower

[unclewebb]

JohnZS: Your sensors are still near the top of my list of bizarre sensors so thanks for posting that. I originally came up with the idea that some of these sensors move along different slopes. At the IDF, Intel jumped on that band wagon and agreed with that theory. With your sensors, slope error might be a minor issue but it doesn't seem to be the major issue. The slopes for all 4 cores are quite similar and none of your sensors are sticking.

Your data would lead one to believe that during calibration by Intel, TJMax may only be accurate to something like plus or minus 5C. The 45nm Atom sensors have Intel documented error amounts similar to that at TJMax but my contact says that the 45nm Core 2 Duo based sensors are not like that at all. After looking at your data, I guess I'd disagree. It's too bad that Intel has decided not to share the amount of error at TJMax with enthusiasts.

I recommend using Prime95 Small FFTs when testing but I tried using "Maximum Heat and Power Consumption" like you did and the results were very similar to Small FFTs. On Max Heat, the power consumption is very consistent at each level just like Small FFTs is. The maximum power consumption at the wall was a few watts less so the peak temperature was about 2C less when I tested but each step down in power was unique and very consistent. Linpack testing creates more heat but it doesn't have the consistency that Prime95 has.

I'm still looking at data at the moment. When testing, you'll get the best results if you make sure that there isn't a lot of junk running in the background. Vista has a lot of background baggage like SuperFetch you need to watch out for.

[killerhz]

How are my sensors? I am just starting to follow this thread and learn about this topic.

[randomizer]

I am just starting to follow this thread and learn about this topic.

The search for truth is addictive I tell you. It's also difficult when you get so much conflicting information like this:

Your data would lead one to believe that during calibration by Intel, TJMax may only be accurate to something like plus or minus 5C. The 45nm Atom sensors have Intel documented error amounts similar to that at TJMax but my contact says that the 45nm Core 2 Duo based sensors are not like that at all. After looking at your data, I guess I'd disagree. It's too bad that Intel has decided not to share the amount of error at TJMax with enthusiasts.

[justdrew424]

I'm offering up some screenshots just in case you need more data. My system is a Q6600 @ 3.2 Ghz.

Here is a shot after running the sensor test.


Here is a shot during calibration. Room temp was 23 C.



Drew

[randomizer]

What cooling are you using Drew?

[concretefire]

Justdrew..........I could be wrong but you state you are running a Q6600 @ 3.2. Your screenshots show otherwise. I believe you have your speedstep and C1E in your Bios "ENABLED". These are power saving features. Look at your CPU-Z screen. Your multi is 6 , not 9.
At any rate, the ONLY reason I type this to you is just an FYI. If you don't / didn't know this about your CPU , you might believe you are running 3.2ghz all the time, when clearly you are not. **( I didn't know either )** Just be aware of that as your low idle temps could be very decieving to you while these power saver features are on.

Unclewebb: THANK YOU SIR!!!!! THANK YOU THANK YOU THANK YOU!!!!!. I am finally starting to figure out this whole TJmax thing and this program is the best (my opinion). I do have a question for you though, and a screenshot. This is not a flame question. I promise. I just want to get to the bottom of it. By default in your program, TJmax is set to 100C. I too am running a Q6600 GO and based on what Intel just released I believe *MY* Tjmax should be adjusted to 90C. I have also read some 'this can't be' type of posts, so I am asking you......THE MAN!!!

Before I post my screenies, I have my first ever success overclocking tweaking story thanks to your program. I have my Q6600 @ 3.1 via 344 x 9 (still just in testing mode) (Speedstep off, C1E disabled) Going by your instructions, Prime95 failed on core 0 every time , but did manage to finish the testing. I couldn't figure out why. So I reboot, upped Vcore 2 notches , and tried again. PASSED 3 times.
I'm so happy I could eat ice cream. Please advise me of something else.....by the screenshots I'm near 30C to TJmax with the stock OEM cooler. No lapped anything. (good weekend project coming up) Is this too close for comfort? Thanks again Uncle Webb

[justdrew424]

Cooling is Xig S-1283.

Yes the stepping is on. While Prime 95 is running it shows the 9x multiplier. I only use the idle temps for calibration. I only worry about max temps when under load.

Later
Drew

[concretefire]

Cooling is Xig S-1283.

Yes the stepping is on. While Prime 95 is running it shows the 9x multiplier. I only use the idle temps for calibration. I only worry about max temps when under load.

Later
Drew

That's cool. When I first saw my CPU-Z reporting a 6 multi I was like . I had no clue what was going on. So I was just trying to be helpful to you in case you didn't know. What do you think about me being 33c from Tjmax?

[justdrew424]

I don't think that is all that bad. At the moment I am OC'ed to 3.6 and running a folding SMP client. My cores at sitting at 31 - 38 from TjMax.

[unclewebb]

The new Test Sensor option in RealTemp is providing a lot of new information about what these sensors do. It's kind of like a Rubic's Cube. Hard to walk away when the puzzle hasn't been solved yet. Welcome to all the new users joining this thread. We always like to believe that we're getting closer to the real core temperature.

concretefire: When overclocking, the low 30s to TJMax is about as close as I like to get. You don't have to get too scientific about this number though. If your processor is too hot then it will either error out during a Prime test or it will simply re-boot. If you are Prime stable then you are good to go. If you are Prime stable but not game stable then I would look at what temps your GPU is running at. GPU manufacturers seem to be pushing the temperature envelope these days with some hot chips and barely adequate cooling solutions. GPU-Z is a nice little app for that.

TJMax for the Q6600 is still being debated. The information Intel has recently released goes against what an IR thermometer says for many 65nm processors. I originally assumed TJMax=95C because that's what an IR thermometer reads when DTS=0. Based on rge's testing here at XS, it's very likely that when the IHS surface temperature is 95C, the actual core temperature has to be slightly higher or about 100C. There is more heat dissipation through the IHS than I originally assumed. RealTemp will be using TJMax=100C for the Q6600 G0 until I see some testing from Intel that shows otherwise.

The 90C number that Intel says we should be using for the G0 just doesn't seem possible.

[angelreaper]

[unclewebb]

angelreaper: Pretty obvious that your Core 3 sensor is stuck. The other 3 sensors look good.

I just found some pics of early testing of my Q6600 G0.
When the IHS was reading this:



RealTemp using TJMax=95C would read the same temperature.



To maintain an IHS surface temperature of 60C, it's reasonable to assume that the hottest spot on the core is about 5C hotter or about 65C. To get RealTemp to display that number you need to use TJMax=100C. If I used TJMax=90C like Intel recommends then the core temperature would only be reported at 55C. There's no way the IHS heat spreader can be hotter than the source of the heat.

Edit: Those pictures were taken 2 seconds apart.

[concretefire]

Ok Unclewebb. I'll move it back to 100c until you say different for my Q6600 GO. Thanks for the clarification. Two more quick questions if you have the time.

1) Could you explain in a nutshell what the "movement" of the cores represents? (just quick laymens terms)

2) Back to the TJmax thing......this weekend I will be lapping the CPU to remove the IHS. In your opinion , if it was your chip, would make any adjustments to the TJmax setting because of this?
( I realize intel says 90C, you say 100C, so i'm thinking lapped maybe 95c as a "middle ground". Just a thought. Wanted your input)

[burebista]

Man, try to forget TJMax. It doesn't matter for throttling (put whatever value you like, distance to TJMax will always be the same).
All that matter is distance to that pesky TJMax (which is read straight from registry). Keep it >25-30 along with a stable rig and be happy.