RealTemp General Discussion

[sparkyork]

ty unclewebb, indeed my c1 state looks like this

 

[unclewebb]

I'm not sure if you have a C States bios setting that might let your CPU go into one of the deeper sleep states to conserve a little power and reduce the idle temperature a little. To be honest, it probably won't make any significant difference even if it can go from C1 to C3 or deeper.

Is your computer case open or closed and how is the air flow. When testing I often times open up my case for consistency. What GPU are you using? Trapped heat can increase the core temperature.

 

[sparkyork]

i think my pc does go into a sleep state when left for a while (you cant even tell its switched on)

im gonna restart now and enable any power saving things and see what happens

my case is open and im running a ati radeon 5770, akasa evo cooler which is supposed to be pretty good

id like to just know what temps other 9300 users are getting or were getting before going newer lol

 

[unclewebb]

I'm using a 5770 too and like how cool it runs.

These sensors have more than +/- 10C of error in them at idle. If the CPU is actually at 40C then one sensor, like yours, might report that CPU as 50C and another CPU might report that exact same temperature as 30C. Comparing temperatures to other users using these sensors is pointless. They are just not that accurate.

They can only be compared to themselves. If you reboot and change something and the sensor reports a cooler temperature then you've probably done good as long as this change shows a repeatable difference with some A/B/A type testing where you can prove your results.

 

[bouchigo]

Unclewebb, or anyone else who can help,

I need help interpreting this info.

The ambient temp. around my pc was measured at 20 degrees Celcius, and the temp. inside my computer case was measured at 25 degrees Celcius.

As I am sure you already know, TJmax for the i7-860 is 99 degrees Celcius.

These are the results I got (I didn't make any adjustments for idle calibration):

 

[unclewebb]

The sensors Intel uses are not 100% accurate, especially at idle. TJ Target is 99C but actual TJMax is unknown. There is a certain amount of variation in this number and Intel has never documented how much. The amount of error at TJMax also varies from one CPU line of theirs to the next.

These new Core i7 CPU sensors also have slope error. That means as you get further and further away from the point where Intel calibrates them, they can either start reading a little too high or a little too low. This happens to different cores on the same CPU. No one knows which cores are correct or what cores are closest to correct or what cores are completely wrong. It varies from one CPU to the next.

There are too many variables and too much inside information not documented by Intel.

 

[bouchigo]

Do you have any suggested adjustments I can make? That 37C on core 0 seems a little too high to me based on ambient temp. and temp. inside case.

 

[unclewebb]

It's harder to come up with a meaningful calibration for these because I don't have any data about them. You also have 3 of the 4 cores being able to go to sleep at idle where the core voltage gets reduced to next to nothing and can run quite cool.

Your screen shot shows all 4 cores reporting the exact same minimum temperature and the minimum looks very reasonable considering your room temperature. If core 0 is mostly awake and the other 3 are mostly asleep then the difference doesn't look too unusual. If you use a negative calibration factor to reduce the reported core temperature of core 0 then that is going to drop its Minimum temperature way below the other cores. That's not right.

I'd just leave it. If you don't like this one reporting too high then ignore it at idle. There's not enough known about how these CPUs work internally to decide how accurate the sensors are. Intel says their sensors are not designed for accurate temperature reporting and I agree. As long as your CPU is not thermal throttling or running unstable then you don't have to worry about the core temperature anyhow. It's just a number.

 

[bouchigo]

Well, when you put it that way, I guess it does make sense since core 0 is the most active. I will just ignore it for now as you've advised and hopefully in time (hopefully not too much time) more info. will come about for these processors to get more accurate readings.

Thanks

 

[unclewebb]

You can also try using ThrottleStop. It gives you an accurate look at what each thread is up to in terms of load and the average multiplier.

ThrottleStop 2.00
http://www.fileden.com/files/2008/3/3/1794507/ThrottleStop.zip

 

[bouchigo]

Thanks, I will give that a try.

 

[Sergey Lepilov (XBT)]

unclewebb
Hi!

1) Thanks for you programm!

2) The plugin for the RivaTuner does not work with the last RivaTuner and RealTemp 3.59 on the Gulftown. Have you a new version of this plugin?

3) Why the temperatures of the RealTemp and the CPU-Tweaker monitoring are not equal?



Who it is right?

Thanks!

 

[burebista]

1. Me too.
2. Plugin is only for non-Gulftown Intel CPU's
3. First of all TJMax (TJTarget in fact) is 101 not 100 and second I'm sure that CPU-Tweaker shuffle cores between reboots.

Who is right? There are no questions, for Intel CPUs RealTemp is the best monitoring software.

 

[unclewebb]

You are welcome. The RealTemp - RivaTuner plugin has not been updated in a long, long time so it does not correctly support the new Gulftown 6 core CPUs. I will think about updating this in the near future.

When lightly loaded, the core temperature can change very rapidly a few degrees up and down, especially on these newer CPUs that are able to put cores into a low power sleep state. The best program to test with is Prime95 with the Small FFTs option. That puts a very consistent load on each thread so if you run that then it will be a lot easier to compare two different monitoring programs. Can you post a screen shot of that at full load after the CPU temperature has stabilized?

RealTemp also uses thread ordering so it associates each thread with the correct core. Most software doesn't bother doing this so what RealTemp reports for Core 2, another program might report that data as Core 4. In the RealTemp Settings window at the bottom is a number called APIC ID. Can you see what that shows when you are doing your testing. That can change each time your computer boots up but is usually consistent from one day to the next. It can also change when you flash the bios. Knowing this information will make it easier for me to compare both programs when they are at full load.

As burebista mentioned, TJMax is written into each core of the CPU now. I know RealTemp reads that correctly but I'm not sure about other software.

 

[unclewebb]

Sergey Lepilov (XBT): I've updated the RealTemp - RivaTuner plugin for better Gulftown support but the problem is that I don't have a Gulftown to test this on.

Can you explain to me what problem you are having with it? Do any of the graphs get displayed in RivaTuner? Are some missing? Does it not work at all?

If you can show me exactly what the problem is then maybe I can fix it. Thanks.

This updated version should correctly support the QX Core 2 CPUs now.

RealTemp - RivaTuner PlugIn - 3.5.9.0
http://www.fileden.com/files/2008/3/3/1794507/RTCore.zip

 

[toohot]

Hi Uncle.Its been some time since I have updated Real temp current version 3.58.1 and working great.Should I update version to 3.58.4 or later


Case details are coolmaster cm690. front 120mm fan 1200rpm/rear fan 120 mm 1200rpm/left fan 140mm 1200rpm. cpu fan Noctua Nh-u12p se1366 performance cooler / ati radion hd 5770 graphic card/gigabyte ga-x58a-ud5 mb/ram G.Skill Trident f3-12800clt6t-6gbtd/win 7 64 pro/hdd x3
wd 1TB(wd10ears07200rpm 64mb satta 11 green power/cpu intel i7 920/thermaltake toughpower Qfan 850 watt

 

[unclewebb]

I always like upgrading to new software. It makes me feel good even if the old stuff is working fine. It's probably a sickness.

RealTemp 3.59
http://www.fileden.com/files/2008/3/3/1794507/RealTempBeta.zip

I try to fix more things than I break with each new version. The recent fixes have been mostly for Core 2 Extreme CPUs so you might not notice any big difference besides a new number.

 

[Vhozard]

I just wanna say I had the exact same problem with the Reset-button (using ATI here too), but the version you uploaded fixed it.

Thanks!

 

[unclewebb]

If you have an ATI GPU you can try adding GPU=2 to the RealTemp.ini file.

 

[customize_your_sleep]

im running a q9550 in a coolermaster HAF full tower runing stock heat sink. ive had it for about 6 months and probly put about 800 hours on it no overclocking but i find it odd that suddenly after i got 3.40 my cpu temps are now at 45 c idle and 3.00 said 32-34 befor and it also says 45 now aswell as core temp w/room temp at 71. was there a update to cpu temp monitors or did my heat sink just crap out on me.

 

[Mussels]

im running a q9550 in a coolermaster HAF full tower runing stock heat sink. ive had it for about 6 months and probly put about 800 hours on it no overclocking but i find it odd that suddenly after i got 3.40 my cpu temps are now at 45 c idle and 3.00 said 32-34 befor and it also says 45 now aswell as core temp w/room temp at 71. was there a update to cpu temp monitors or did my heat sink just crap out on me.

check your TJmax setting. it may be defaulting wrong.

 

[customize_your_sleep]

i dont know what the tjmax settings supposed to be at there at 100 i think

here an img of temps and settings

http://i562.photobucket.com/albums/ss61/Customizeyoursleep/Untitled.jpg

 

[Mussels]

i dont know what the tjmax settings supposed to be at there at 100 i think

here an img of temps and settings

http://i562.photobucket.com/albums/ss61/Customizeyoursleep/Untitled.jpg

i dont think its supposed to be at 100. please read up on TJmax and try and find the value yourself.

 

[customize_your_sleep]

ok thx will do or try to do anyways

 

[customize_your_sleep]

im wondering if i chage the tjmax in real temp will it have a physical effect on my cpu?

http://www.tomshardware.com/news/intel-dts-specs,6517.html

i tried looking into it but i can seem to get a understanding of how tj max exactly works

so that was the best i could find.and i still have no clue whats going on or how to change tjmax or if it actualy causes the cpu to reduce power to drop heat.