Real Temp - New temp program for Intel Core processors

[Arthur Dent]

Thanks for the recomendation but is CPUID reliable?

Core temps are +5c compared to RT, and just before i snapped this image the current cpu temp was 210c/409f for quite a while and actually has returned to that setting since i took the shot!!

please also note that min cpu temp shows 15c in a room at 25c constant and im only on air cooling!!

i think if im running 210c on idle, then running prime95 at full load may be too much for my little old Q6600

Having said the above, could PWM also be misread in CPUID?

Once again thanks for the help.

[Arthur Dent]

Well i guess im not so cautious as i thought....i decided to run prime95 anyway, 210c idle or not.



restarted latest version of CPUID after prime95 already running and was surprised to see minimum cpu temp went down to 12c (10c below ambient temp on air), and current cpu temp down to 17c, a fantastic drop of 190c or so despite the 100% load on it and the cores being in mid 50's ??

am i doing something wrong here ?????

[loonym]

am i doing something wrong here ?????

The only thing I can see you might be doing wrong is trying to rely on unreliable software to read board hardware temps. Pull the sinks off your chipset and redo the TIM, maybe do Ace's boltmod as seen in the IP35-pro thread and put a fan on the mosfets and nb. You'll feel much more confident when overclocking.

[i43]

hi Arthur,
don't bother Yourself with that CPUID HW monitor, this runs as it like, LOL sorry
use RealTemp instead and it will be fine
by my own experience and tests, i suspect that CPUID HW monitor gets stuck on some circumstances, and not only on temperature readings ..
(seen this ..1C ; 179C and 236C myself as well)
... BTW i have seen it also reporting CPU vCore something 3.33 volts on QX9650 .. and this is wrong
my processor is still alive .. so far

( i don't know why this happens, but i have seen similar ''wrong'' readings with Asus PCprobeII
with different processors on on different asus boards ( p5k3-dlx and MaximusExtreme)

[Arthur Dent]

Thanks all.

i do use RT for all my core/cpu readings as its the best i have ever seen.

was trying to get some idea how NB & PWM etc were doing - wish RT could show PWM.

I intend to do mobo mod you mentioned soon but no access to make device (im disabled) to add fan and cant see anything to buy on web, so im putting the mod off for now i guess.

[loonym]

Thanks all.

i do use RT for all my core/cpu readings as its the best i have ever seen.

was trying to get some idea how NB & PWM etc were doing - wish RT could show PWM.

I intend to do mobo mod you mentioned soon but no access to make device (im disabled) to add fan and cant see anything to buy on web, so im putting the mod off for now i guess.

I use 40mm fans held on with zipties. Be creative

[rge]

am i doing something wrong here ?????

No but either the dummies who wrote your bios did, assuming you have an updated bios, or your mobo sensor chip has too much interference to get accurate cpu temps.

Core temps, ie Real Temp, are obtained by reading output directly from cpu where DTS reads/calibrates transistors functioning as thermal diodes, and there is no worry of electrical interference from mobo or poor bios calibration/support, etc.

CPU temps, the diode between the cores, must be calibrated/supported by your bios. If bios writers did a poor job of cpu calibration...you get erroneous temps. Also since the cpu temps are read by an off cpu sensor on mobo, they are subject to all kinds of electrical interference on mobo.

Though you could try speedfan, and make sure not a software bug on program you are using to read temps, but I doubt it. Load core temps should be as accurate as you can get. cpu temps, idle or load, usually require end user calibration, and good bios support that can predict all the variable end user configs and account for most.

[zorzyk]

Arthur, as from my experience uGuru on abit IP35 PRO BIOS 14 (official) and Q6600/B3 is reliable enough -
- CPU temperature is shown almost exactly the same as RealTemp does
- PWM temperature is shown correctly (tested with my own finger )

No other BIOS (up to 17 beta 4) even come up to BIOS 14 - temperatures are completely screwed: CPU 70*C etc...
So try to check different BIOS version and probably it will work for you better.



Edit: needless word deleted

[Arthur Dent]

I do understand what your saying about the bios as my last board was Abit IP35 pro and i did have to backstep bios updates to get anything sensible, 16 was hopeless & 14 seemed solid.

now i got the Abit IP35 pro XE and its not quite the same thing it seems.

I always use realtemp (2.68) and cpu-z (1.45) for monitoring and under motherboard bios it shows version 6.00 PG date 02/29/2008.

any ideas what im missing ?? (apart from the obvious)

[W1zzard]

- CPU temperature is shown almost exactly the same as RealTemp does (sic!)

please read what it means before making a fool out of yourself: http://en.wikipedia.org/wiki/Sic

[Arthur Dent]

ZORZYK just a quick update..

Went to abit site and found i am on BIOS ver 629o10 (also known as ID10), not 629b14 (also known as ID14 or bios version 14) as on IP35 pro (not XE).
XE ver ID11 released only yesterday but cant see why to update unless i go q9450, however i have seen several postings where people have flashed the XE bios onto the regular PRO mobo..

i have not tried yet, but if Abit are still using same flash methodology then XE flash update will refuse to install bios meant for IP35 Pro - hope that helps anyone out there thinking of buying an XE.

[unclewebb]

please read what it means before making a fool out of yourself:

Come on W1zzard, take it easy on my friend Zorzyk. His recent input about core ordering has been very valuable to RealTemp and I'm hoping that we can continue to work together by PM. He may not be an English major but I can say that his testing procedures and thorough presentation of data are second to none.

Arthur Dent: Thanks for your post. Looks like RealTemp agrees with HWMonitor when it comes to the core temperature order. RealTemp goes that extra step and puts your core2 in its proper place in the GUI. Does anyone else have a HWMonitor / RealTemp comparison of core order for a Quad that they can post? (45nm or 65nm) I don't agree with most of the other data that comes from HWMonitor but at least if it agrees with RealTemp about core order then it's a keeper! Unfortunately I can't test HWMonitor on my motherboard because it's still hiding all 4 core temperatures.

[rockfella]

I was under the impression that CORE TEMP is considered the best in measuring CORE2DUO temps!

[zorzyk]

Q6600/B3: HWM shows the same core order as RealTemp.

[belial]

quick question, i got a e7200 and p5q pro, the sensor on core1 reads 7 when i press "test sensor" this core runs MUCH hotter then core0, which reads 0, is there any ways to fix this ?

[rockfella]

So what do we have? CORE TEMP or REAL TEMP???????????

[rockfella]

CORE OR RT Guys???????????????????????????????

[perso]

Uncle, here are my results.
Seems prime does not work only on cpu0 , all four temps
increase .
(Sorry for my ugly english)

[rockfella]

When i run my E2140 @ stock volts CORE TEMP (100 TJuntion) i get a whopping 42 deg, where as BIOS shows 36 deg, everest shows 32! I am confused, i have neevr tried REAL TEMP. Anyone????

[balewolf]

Should i trust this program or coretemp i got some weird temps i have a 1.20 vid q6600 and testing it at 1600mhz and 1.09 vcore my room temp is 18-20c but it idles at 32c same idle temp as when i run 3700mhz with 1.31 vcore. When i run 3700 i load up in 70s with prime 95 and only using 1.31 vcore. My tuniq tower hardly even gets warm to the touch and i have remounted it 20+ times always same temps or worse.

[loonym]

When i run my E2140 @ stock volts CORE TEMP (100 TJuntion) i get a whopping 42 deg, where as BIOS shows 36 deg, everest shows 32! I am confused, i have neevr tried REAL TEMP. Anyone????

Why not try it and judge for yourself? It's extremely user friendly and the calibration and other features make it the best imo. In the end it really comes down to personal preference and only you can decide that.

[unclewebb]

CORE OR RT Guys???????????????????????????????

Here's a test I did on my E2160 which is similar to your processor.
http://www.xtremesystems.org/forums/...&postcount=423

You get to decide what program is giving you an honest temperature.

Edit: I found this info for a Q6600.
http://www.xtremesystems.org/forums/...&postcount=568

The E8400 test is pretty thorough:
http://www.xtremesystems.org/forums/...&postcount=573

If you see any real world testing from the competition then please post that info. I haven't seen anything from them yet.

balewolf: Show us a screen shot of your Q6600 running Prime95 and what RealTemp shows for it. Include some details like your room temperature and whether your case is open or closed, etc. Sticking sensors is much rarer in the previous 65nm processors but it does happen. Your sensors might be sticking down low.

I was under the impression that CORE TEMP is considered the best in measuring CORE2DUO temps!

I think it was too, before RealTemp came along that is and proved that CoreTemp might have some nice features but accuracy isn't necessarily one of them.

[balewolf]

amb is 18-20c and my antec 900 is closed and all fans on high currenty im running 1600mhz but my idle temps are the same when i run 3700 with 1.31 vcore.

[unclewebb]

baleworf: Your temps don't look too out of the ordinary. Here's how my Q6600 looks like at 1600 MHz. The room temp is 21C.



I need to use calibration factors of approximately +1,+1,+2,+1 so that my idle temps are pretty well balanced. Since that screen shot I've introduced the ability to use decimal calibration factors which gets things a little closer yet.

I don't believe you have sticking sensors. A little unbalanced maybe but calibration can really help that out. Try posting two screen shots at 1600 and at 3700 MHz. Show RealTemp and CPU-Z to report core voltage. With a well ventilated case there won't be a huge difference in temps but you should see something if your testing methods and room temperature are consistent. I like doing this kind of testing with the case open. No matter how good your case is, I usually see lower temps with an open case.

[unclewebb]

Here's another test that I just did similar to your screen shot. I ran Prime small FFTs (8K) about half way through the first test. Core voltage and MHz were similar to your numbers and I'm using a Tuniq Tower on high with the case open. Room temperature is 22C today.



My numbers don't look a lot different than your numbers. Your core2 and core3 are reading too low and definitely need a calibration boost. Let your computer sit for 5 minutes without any junk in the background running, heating up your CPU. Let core0 be your baseline and adjust the other 3 cores so the idle temps are equal. Do this at 1600 MHz and 1.10 volts and your overall accuracy will be excellent. You might need a tiny adjustment after this but not much.