C2D temp question; Discrepencies among monitoring programs

[BryanHilt]

I have a question about Core 2 Duo temperatures. I have thoroughly read (and understand about ninety percent of) the stickied guide on the subject, before anyone suggests I start there.

I just successfully built my first from-scratch PC, and am interested in overclocking the processor a bit, but the temperature readings I am getting from various sources are confusing, and I want to make sure I'm not cooking my system before I do anything stupid.

I'm using a:

Gigabyte GA-965P-DS3 rev 1.3
Core 2 Duo E4300
2 X 1 G.Skill 6400 RAM

I'm using the stock CPU cooler and my case is an Antec 180b with all three fans running on high. The readings I am about to give are for a system that is not currently overclocked in anyway.

So anyway, when I run TAT, both cores idle around 50c, full load around 70c, which according to the stickied guide is pretty high, especially considering the fact that nothing has been tweaked.

However, when I run Core Temp and SpeedFan, they have almost identical readings that are a good 15c - 20c below what TAT is showing: both cores idle around 31c, full load (with TAT) around 57c. Also, the temperature in SpeedFan that corresponds to the Tcase idles around 22c, a few degrees higher than the temperature my BIOS displays on startup, which seems to fit.

What gives? Do I trust TAT, which suggests that, for whatever reason, I am already pushing it temperature-wise? Or do I trust Core Temp and SpeedFan, which both suggest that I've got plenty of room, at least in terms of temperature, to overclock my CPU?

One last thing that is probably important: after first seeing even higher temperatures in TAT, I opened up my box and discovered my CPU cooler was visibly loose. I reseated it, and my temperatures dipped accordingly. I assume for this reason that any problems I am having don't have to do with re-seating my cooler.

Thanks a ton, and in advance, to anybody who responds to this.

 

[ramsfan13]

This will be a huge debate, but I would trust coretemp .94 because I believe the E4300's have a Tjunction of 85, not 100.

 

[orangegator]

I have read a great deal about this too. I believe ramsfan13 is correct. Speedfan, Coretemp .94, and Everest are correct. TAT and Cortemp .95 read 15C too high. I think this just applies to the E4X00 chips. They should read correctly for the E6XXX chips and all temps should be pretty close. Though I'm sure some people will argue otherwise.

 

[aLdaRiS]

HERE HERE!!! ILL ARGUE OTHERWISE
the reason i say e4300 tjuc is 100c is... because a lot of ppl were using the 85c tjuc n supposedly they were idling BELOW ambient temperatures...on air which is... impossible.., a quick test u could do is... underclock ur cpu, have the lowets vcore and if it idles below "ambient temperature" then you'll kno u got a 100c tjuc...

 

[orangegator]

I haven't heard of idling below ambient. But thermocouples can be a couples degrees off, and if someone's using a great heatsink (ie Ultra 120), then they could be very close to ambient +-error. It just seems odd that sooo may people get ridiculously high idle temp readings with TAT and Coretemp .95 I read a lot about this. Here is the thread that finally convinced me.

http://www.overclock.net/intel-cpus/192672-my-talk-w-intel-about-tat-4.html

Just curious, what are your idle and load readings with the computer in your sig?

 

[13thmonkey]

FYI, I'm running an E6600 in a P180B, and have all of the fans on low (2 in the upper section). By TAT / Speed Fan high 30's idle, and mid 50's load, with a slow fan (stock) speed of 1200rpm.. So even given a 15C offset I'd say that you are a little bit warm.

Are you 100% sure that you have the HSF securely clipped in place? Its a very common problem, that unfortunately can only be resolved by taking the mobo out of the case, (unless you like the feel of your board bending).

 

[Lionhardt]

Wait, i have almost the same setup, except that i use an arctic cooling freezer 7 pro w/ an e4300. I use coretemp .95, since TAT wont work, and i get 59 C on full load. should i try coretemp .94? are those better?

 

[orangegator]

FYI, I'm running an E6600 in a P180B, and have all of the fans on low (2 in the upper section). By TAT / Speed Fan high 30's idle, and mid 50's load, with a slow fan (stock) speed of 1200rpm.. So even given a 15C offset I'd say that you are a little bit warm.

Are you 100% sure that you have the HSF securely clipped in place? Its a very common problem, that unfortunately can only be resolved by taking the mobo out of the case, (unless you like the feel of your board bending).

Actually, his temps seem reasonable using the stock hsf. The E4X00 runs relatively warmer than E6XXX. I think an E4300 overclocked to the E6600 stock of 2.4 would run hotter than the E6600. So both at their stock settings running at similar temperatures seems reasonable.

 

[aLdaRiS]

44c idle in the summer, goes up to 70c, ive reseated my HS/F 6 times.

 

[orangegator]

44c idle in the summer, goes up to 70c, ive reseated my HS/F 6 times.

Is that measured with TAT and Coretemp .95? That seems high (by about 15C ha ha) for using a Zalman 9700 and only 3.0Ghz and 1.290V. Did you read the link I posted? What did you think of it?

 

[aLdaRiS]

44c idle in the summer, goes up to 70c, ive reseated my HS/F 6 times.

Is that measured with TAT and Coretemp .95? That seems high (by about 15C ha ha) for using a Zalman 9700 and only 3.0Ghz and 1.290V. Did you read the link I posted? What did you think of it?

havent read it yet, i will later tho ;p nah i think its because of the horrible concaveness of my processor (imma lap it 2day or 2morro) and yeah... i think it doesnt have such good contact with my hs/f

 

[13thmonkey]

Didn't know that the 4's were warmer than the 6's.

I'm willing to be corrected on this too...

I've just upgraded from an asrock 775dualvista with my E6600 to a P5B deluxe, and idle temps are up about +5C, even though the stock fan RPM is up from 1140 to 1215 (not enough to make much of a difference I know). I had to reseat my HSF, and looking at the mechanisms involved it would not suprise me to find out that the removal-refit cycle would leave the HSF with a slightly lower contact pressure due to rounding of edges on the plastic legs etc. I refitted probably 3-4 times in the end.

Just a thought..

 

[Gorgon]

I'm a DTS convert. I don't care about absolute temperatures anymore since there's really no way to know with exact certainty what the real temp is. Intel is doing away with every sensor except for the DTS sensors. These sensors will give you a VERY acurate reading regarding how many degrees are your cores from reaching tcc (thermal control circuit) activation temperature.

This seems to be the only flawless way of knowing (relative) core temps. Even better, its a universal way of comparison between diferent processor temps in different PCs.

Check out this thread and learn about it. Read CAREFULLY and become a convert!

http://www.xtremesystems.org/forums/showthread.php?t=131008

 

[goldragon_70]

I've been having a bit of a problem too when running the temp programs. Core .95 and .94 will say I'm at 54oC Ideal and 70oC at full load (after several hours on Orthos), and so will RightMark's CPU thingy (sorry not at my computer and can't remember the full name), but I finally got the newest speed fan to work on my computer, and it is show exactly 15oC less in every temp reading over the other two. So my Ideal is about 39oC at Ideal, and 55oC full load after several hours with Orthos. TAT gives me an error, so I can't get those results. Which should I hold more true?

Eather way the temps are high, and I know the HSF is seated right with just enough thermal paste, b/c I've mounted it carefully more then a few times to see if I can get a better fit. I put a box fan at an angle putting are into the case and got a 3 to 4 oC drop (still not good though).

 

[BryanHilt]

Wow, I didn't realize there was so much debate over this. OrangeGator, thanks for the link, I'd not yet read anything about the whole Tjunction 85 vs 100 debate, and the thread has me convinced that you are right, and that SpeedFan/Core Temp .94 are displaying correct temperatures. Which is pretty much awesome.

 

[jonny_ftm]

Sad to see youre going the wrong way.
I used the most recommended way to see the accuracy of sensors.
Ambiant case temperature measured with a digital sensor: 25°
I have a P5B deluxe and E4300.
Set the FSB to 6x200MHz and voltage to 1.10v, on idle:

ASUS Probe reading Tcase: 13° (so probably 15°+/-3 below real temperature) : So ASUS Probe = Bios are 15° lower and wrong

CoreTemp 0.94: 20°C, so also wrong, false and erronus
CoreTemp 0.95 and TAT: 35°C, that is the recommended Tc (ambiant in my case) + 10° to 15°C

Conclusion: In my case, CoreTemp 0.95 and TAT are true

Most E4300 are Tj = 100°C

Now, if you like to believe what comforts you instead of the truth, you're free. But before going the wrong way, just lower speed and voltage, idle your CPU, and check your sensors: you'll be scared to see how hot you're running

 

[jonny_ftm]

About TAT and coretemp 0.95 3-5° discrepancy, is that because TAT was built for mobile CPU, without a heat spreader. So CoreTemp actually corrects that by adding the 3-5° delta due to the presence of a heatspreader

About reseating your heatsink, I reseated mine 4 times, without a single drop in temperature. At 4th attempt, I changed the mode that I applyed the thermal compound (so stopped using the grain of rice in the center technic and used the virtual layer one + more compound than a grain in center) and it got me a 10°C fall on load at lower CPU fan speed

So yes, unproperly seated heatsinks are the major cause of high temperatures. You have to try many times, but also different ways to find what suites you.

Also, don't believe all that people say. The threads linked above to show us that TAT is wrong for example. One of them was monitoring his CPU to throttle. Well, E4300 chips with a Tj of 100°, will throttle at that temperature and not 85 or 93°C. The guy was even thinking that TAT is responsible for the throttling mechanism (just to show people ignorance) while it is internally built into the CPU/MB bios couple. Also, if at 101° it didn't throttle, don't forget that these are not absolute temperature readings. They are only a delta to a fixed Tj, that with ambiant temperatures/cooling/sensors accuracy, can give you up to a +/-10°C

And at 110° TAT, I'm sure your CPU won't throttle, but rather Windows will crash / reboot before

Also, if you read the forums, E4300 chips, seeing the coretemp results of users, are probably all Tj 100°, except the first engenniering samples probably

So, every one should test his idle at close to ambiant temperature to calibrate his sensors. After that, come and give us consistent results, and not postulaions like: "coretemp 0.94 is right while newer coretemp 0.95 which added correct E4300 Tj detection is wrong", just because it suites your needs: a miraculous 15° drop.