PDA

View Full Version : QX9770 Temperature


Silllllll
05-08-2008, 02:17 PM
Hello everyone, I'm new to this forum and I thought this was a good place to ask for help regarding my problem.
I recently got a QX9770 cpu. I use this on a Gigabyte GA-X48T-DQ6 mainboard.
However I seem to have very strange temperature readings.
I'm using a Zerotherm Nirvana cooler with artic silver 5 thermal paste. I have remounted the heatsink several times with no improvement.

IDLE TEMPERATURES:
Bios says the temperature is 33C
PC Health monitor(which is a program delivered with the mainboard) 24C
Real Temp 2.5 : Core 0/1 41/41 2/3 32/35C
HWmonitor : Core 0/1 51/51 Core 2/3 42/45 ( The TMPIN1 is 24C which moves in sync with the PC Health monitor cpu temperature)
Core Temp : Core 0/1 51/51 Core 2/3 42/45
Speedfan : Core 0/1 46/46 Core2/3 37/38
Everest : Core 0/1 51/51 Core 2/3 42/44

The heatsink feels cold, I have tried 3 different coolers. I tried my old intel stock cooler for e6700 which resulted in a 48C idle in the bios and 70C+ when running Prime 95

Thermalright IFX-14 was 50C idle in bios (no fans attached) I could potentially blaim this to bad mounting. However seeying how I'm positive the Zerotherm is mounted correctly, this situation confuses me, if anyone could give me any info and help it would be appreciated.

Cutthroat
05-08-2008, 03:03 PM
Real Temp and Speedfan are wrong, the others are correct. 45nm quads run hot.

Silllllll
05-08-2008, 03:10 PM
Thanks for the reply, I was under the impression tho that 45nm were suppose to run less hot then the 65nm ones. On the intel website it says the max temp for a QX9770 should be 55C
Any additional feedback from you and others is more then welcome

zsdersw
05-08-2008, 03:45 PM
Originally posted by: Silllllll
Thanks for the reply, I was under the impression tho that 45nm were suppose to run less hot then the 65nm ones.

They do. :)

Silllllll
05-08-2008, 03:47 PM
This really boggles my mind then :/

Cutthroat
05-08-2008, 03:56 PM
Originally posted by: Silllllll
Thanks for the reply, I was under the impression tho that 45nm were suppose to run less hot then the 65nm ones. On the intel website it says the max temp for a QX9770 should be 55C
Any additional feedback from you and others is more then welcome

Inside your CPU there are 5 thermal sensors, one under each core, and one in the center of the die. The core temps will likely be ~15C higher than the center of the die. Intel is likely stating the CPU temp (Tcase). Around here most people would say 70C is the highest you want your cores to get for 24/7 usage.

Even with the best air cooler at stock settings I'd be surprised if somebody could keep their quad cores below 55C at load. Tj max on my Q9450 is 105C according to Coretemp, and I believe Coretemp is correct.

My temps, 100% load F@H, 27C ambient:

CPU: 53
Core1: 66
Core2: 66
Core3: 65
Core4: 63

Idontcare
05-08-2008, 04:05 PM
Originally posted by: Silllllll
Thanks for the reply, I was under the impression tho that 45nm were suppose to run less hot then the 65nm ones. On the intel website it says the max temp for a QX9770 should be 55C
Any additional feedback from you and others is more then welcome

You are likely confusing the terms "power consumption" and "temperature".

45nm chips will consume less power than their 65nm analogs. But that doesn't mean their temperatures will be less because temperature is "power density"...a smaller area (smaller chip) dissipating the same power as a larger area (larger chip) will have a higher temperature.

Silllllll
05-08-2008, 04:23 PM
Oh wow I did not realise that, thanks alot for clearing this up. Question tho, if the 45nm run around 50C idle, how are people able to overclock them to 4.0ghz without damaging the cpu? Or do they only do this to show off and dial it back to a more accepteble amount?

Also Cutthroat you said you would be surprised if somebody could keep their quad cores below 55C at load. what would be a "normal temp" on idle?

bryanW1995
05-08-2008, 04:44 PM
Originally posted by: Cutthroat
Real Temp and Speedfan are wrong, the others are correct. 45nm quads run hot.

sorry, you're wrong. real temp and coretemp use the thermal diode inside each individual core to measure temps. so does the latest version of speedfan. however, coretemp believes that tjmax is 105c, speedfan thinks that tjmax is 100c, and realtemp thinks that it is 95c. unclewebb on the xs forums has a LONG thread about this which I'll link to when I'm done, but suffice to say that I believe his program (realtemp) is more likely to be accurate than coretemp or speedfan.

having said all of that, core temps don't mean squat. distance to tjmax is all that matters. change coretemp to display distance to tjmax and it agrees with realtemp (b/c they get their info from the same place). keep your distance to tjmax over 20c and you're good, keep it over 30c and you're golden (all temps at load of course, idle temps are notoriously unreliable and meaningless).

http://www.xtremesystems.org/f...owthread.php?p=2809778 (http://www.xtremesystems.org/forums/showthread.php?p=2809778)

http://www.techpowerup.com/realtemp/

bryanW1995
05-08-2008, 04:51 PM
Originally posted by: Silllllll
Oh wow I did not realise that, thanks alot for clearing this up. Question tho, if the 45nm run around 50C idle, how are people able to overclock them to 4.0ghz without damaging the cpu? Or do they only do this to show off and dial it back to a more accepteble amount?

Also Cutthroat you said you would be surprised if somebody could keep their quad cores below 55C at load. what would be a "normal temp" on idle?

my hsf's on both my Q9450 and x3350 (xeon equivalent to q9450) are never hot to the touch. also, perhaps more telling, my Q9450 @ 3.6 doesn't seem to increase my room temp any more than my e6750 did. These things do NOT run hot, I very strongly believe that unclewebb has got this figured out (at least as well as ****ing intel will allow anybody not on an nda to figure this out).

ps: coretemp now allows you to introduce an offset. use a -10c offset in coretemp if you are just determined to have your temp monitor display "temp" instead of "distance to tjmax". Just remember that the distance to tjmax is more important. As I've told several posters here recently, it doesn't matter if my Q9450 is at 65c or 75c, it's still 30c below tjmax. That's just as good as running a Q6600 at 65c at 100% load, which at 3.6 on air is VERY difficult to do.

imported_ST
05-08-2008, 04:55 PM
Originally posted by: Idontcare
Originally posted by: Silllllll
Thanks for the reply, I was under the impression tho that 45nm were suppose to run less hot then the 65nm ones. On the intel website it says the max temp for a QX9770 should be 55C
Any additional feedback from you and others is more then welcome

You are likely confusing the terms "power consumption" and "temperature".

45nm chips will consume less power than their 65nm analogs. But that doesn't mean their temperatures will be less because temperature is "power density"...a smaller area (smaller chip) dissipating the same power as a larger area (larger chip) will have a higher temperature.

temperature is still a function of power dissapation / consumption and with the 45nm utilizing less current and voltage, this will in turn lower temps too.

Cutthroat
05-08-2008, 06:09 PM
Bryan, the reason I believe Coretemp, Everest, etc. is correct is because my Q9450 before mods ran about 10C hotter than my E6400 did. I could feel the difference in ambient temps that made in my room. Now that I've gotten my temps down to the same as my E6400 was, my ambient temps are back to normal.

But what Bryan said about delta to Tj max is correct, that's all that really matters.

Cutthroat
05-08-2008, 06:29 PM
Originally posted by: Silllllll
Also Cutthroat you said you would be surprised if somebody could keep their quad cores below 55C at load. what would be a "normal temp" on idle?

I can't answer that, too many variables involved. But I can give you my current idle temps. Keep in mind I've lapped my CPU and HSF, and I've removed the CPU clasp, and added washers to the spring screws on my TRUE to increase pressure. I'm using ICD7 for TIM. Ambient temp is 28C, and the side of the case is off.

CPU: 39
Core1: 54
Core2: 54
Core3: 50
Core4: 49

Idontcare
05-08-2008, 06:34 PM
Originally posted by: Silllllll
Also Cutthroat you said you would be surprised if somebody could keep their quad cores below 55C at load. what would be a "normal temp" on idle?

That's obviously an unqualified statement which I'm sure was intended to only be valid within certain constraints (such as "assuming stock HSF", etc).

For myself I have many quads which are overclocked to 3.3GHz and at 100% load across all cores they hit 53-55C. When clocked at stock I don't break above 45C.

That's with Tuniq120. YMMV.

Cutthroat
05-08-2008, 06:40 PM
Originally posted by: Idontcare
Originally posted by: Silllllll
Also Cutthroat you said you would be surprised if somebody could keep their quad cores below 55C at load. what would be a "normal temp" on idle?

That's obviously an unqualified statement which I'm sure was intended to only be valid within certain constraints (such as "assuming stock HSF", etc).

For myself I have many quads which are overclocked to 3.3GHz and at 100% load across all cores they hit 53-55C. When clocked at stock I don't break above 45C.

That's with Tuniq120. YMMV.

Of course I am assuming stock conditions, normal ambients, etc. What I was trying to say is 55C core temp at load, even at stock speed, would be a very good temp.

Idontcare
05-08-2008, 11:20 PM
Originally posted by: Cutthroat
Originally posted by: Idontcare
Originally posted by: Silllllll
Also Cutthroat you said you would be surprised if somebody could keep their quad cores below 55C at load. what would be a "normal temp" on idle?

That's obviously an unqualified statement which I'm sure was intended to only be valid within certain constraints (such as "assuming stock HSF", etc).

For myself I have many quads which are overclocked to 3.3GHz and at 100% load across all cores they hit 53-55C. When clocked at stock I don't break above 45C.

That's with Tuniq120. YMMV.

Of course I am assuming stock conditions, normal ambients, etc. What I was trying to say is 55C core temp at load, even at stock speed, would be a very good temp.

Yeah I didn't see your original post (did a quick eye-scan for it but didn't catch me) but I figured there was more to your statement than what was being quoted.

aigomorla
05-09-2008, 12:07 AM
Originally posted by: Cutthroat
What I was trying to say is 55C core temp at load, even at stock speed, would be a very good temp.

mmm... depends on your cooling.

Ask IDC how he would act if he found out his quad was loading on 55C on his vapo, with no - sign in front.

I bet i can guess 4 words:

See Chicken Little Run! :P

Cutthroat
05-09-2008, 12:51 AM
Originally posted by: aigomorla
Originally posted by: Cutthroat
What I was trying to say is 55C core temp at load, even at stock speed, would be a very good temp.

mmm... depends on your cooling.

Ask IDC how he would act if he found out his quad was loading on 55C on his vapo, with no - sign in front.

I bet i can guess 4 words:

See Chicken Little Run! :P

errr, well I wasn't really thinking of phase-change cooling, but that's a great way to keep your chip running cool.;)