Q9550, One Core Running Hot

D

Deeky

Gawd
Joined
Feb 7, 2008
Messages
982
I've just finished rebuilding my system with Windows 7 and a Q9550. I've begun stressing it to ensure stock clocks are all good and Everest is reporting Core#1 as being as much as 11 degrees hotter than the other three cores under 100% load (Prime95, Large FFTs). The readings?

CPU 40C
Core #1 54C
Core #2 46C
Core #3 43C
Core #4 43C

All settings are stock on a GA-P35-DS3R (F13 BIOS) @ 1.25v before droop, running 2.83GHz and Mushkin Redline running stock 1000mhz 5-5-5-12 @ 2.1v. All voltage settings are otherwise Auto/Normal.

Is this unusual? Should I reseat my heatsink (Zalman 9700)? If this isn't common, and if diagnosing the problem requires a more in depth description of my hardware, please let me know. Thanks!
 
Last edited:
that is odd. You might want to try reseating the sink, but I doubt its a big deal. If those are temps at full load, you're completely safe.

on the q6600 in my system, core 3 & 4 are almost always the same temp while core 1 and 2 are usually within 1-2 degrees of one another. Is that normal? I have no idea :D
 
Yeah, I wouldn't be too concerned if I intended to run at stock, but I'm hoping to at least bump the FSB to 400 for a sane 3.4GHz OC with minimal case noise. If Core #1 is that high at stock, I'm concerned what it will do once I start feeding it more volts. I'd rather not have any core running higher than mid-60's loaded, if I can help it.
 
Last edited:
they say if there is a 10 degree difference could be a problem with a bad sensor or poor contact

re-seat your heatsink.

Did you apply your thermal paste correctly, if using AS5 did you view their website on proper paste application for a quad core?
 
yeah, I hear ya. I'd say try it out for a few days and see how it acts. maybe the thermal compound is heavy where the 1st core is? might want to check the bottom of the heatsink too... could be a shallow area on it
 
I'm expecting a new Auzentech sound card en route from Newegg (to take advantage of my high impedance headphones). When I pop the case to install it, I'll take a few mins to reapply the Arctic Silver and reseat the Zalman. I'll double-check the AS5 page to ensure I'm using proper application skillz. I'll be sure to report back.

Hmm. If it's a bad sensor, not sure what to do about that ...
 
if you plan on RMAing the chip (if you can't figure it out) do NOT lap it
 
No possibility of RMA, unfortunately. The 30-day return policy lapsed long before I had a chance to rebuild the rig. Hopefully it's something I can fix! I'm praying it's something to do with the heatsink or the thermal paste. The fact the Core #2 is reading a little high makes me hopeful. Seems like there's a patch of high heat right on top of Core #1, near enough to impact the readings on Core #2 (#2 is actually getting hotter, now sitting at 49C compared to #1 56C; everything seems to have peaked).
 
Last edited:
Could just be a stuck sensor, or you may have one core that just runs hotter. Either way, you're still well within safe levels. Unless your CPU starts getting close to 100C under load, you have nothing to worry about.
 
Yeah. 71.4C is the "thermal specification" of the Q9550 according to the Intel website, but I've heard that the processor doesn't even begin to throttle until 100C and won't shutdown until 117C or some ridiculous number.

Stuck sensor? Would that be a mobo issue?
 
71.4C is the maximum TCase specification which is meaningless as far as we're concerned since it's measured in a completely different way from how we get our temperature values. 100C is the effective maximum temperature that we have to keep in mind, although no matter what we don't need to worry about thermal damage since the CPU has built-in protections to prevent it from getting hot enough for that to happen.

A stuck sensor would be an issue with the CPU itself, and you wouldn't be able to do anything about it.
 
I remember when I had my first installation, the stock heatsink wasn't seated properly and the core was constantly running at 95 degrees Celsius range and yet the computer acted like it was running fine.

Also, I've installed a new air cooler and now have upgraded to water cooling both instances my 1st core is generally around 5 cores hotter than the rest of the core so I find this pattern to be normal.
 
Ah well, regardless, I'll reseat the sink with fresh thermal paste and see what comes of it. Lapping is tempting ... I'd definitely lap a heatsink without a care, but sanding the heatspreader on a processor might require sterner stuff than I'm made of (even though I'm sure it's perfectly safe and easy to do without damaging the hardware).
 
There's no need for you to lap your CPU. Your temps are perfectly manageable.
 
Not sure if it's notable, but the temp discrepancy continues even during idle periods. Core #1 is the warmest, Core #2 is slightly tepid, and Cores #3 and #4 and nice and frosty.

#1 38C
#2 33C
#3 27C
#4 27C

Here's hoping the reseat changes something. Otherwise, it's a shame I'll never know whether it's simply a stuck sensor or a core that runs hot (though a stuck sensor ... does that make sense if it's effecting the neighboring Core #2?) ...
 
Last edited:
hey deeky, nonetheless could you post if the result changed for you when you reset your heatsink? I would also prefer equal temperature but having a waterblock installed makes the resitting quite annoying if the change in temperature is negligible.
 
I have the same problem though I wouldn’t call it a problem, if you google it up you will find a lot of people complaining about it, after going through lots of them I concluded that it’s ok and been using it for more than a year now (@3.8), try to see what computronix said about it in his core 2 duo/quad temperature guide.
edit: found it
 
Last edited:
My q9550 does it to, and I just ignore it now. As mentioned above, it's not much of a problem, once you learn to ignore it. Games all work great, it passes IBT, so whatever.

I am running mine at 3.66Ghz and have never had any issues. It was always hotter than my e6850, and even OC'd it's been rock solid. (e6850 was 25C idle, this is 40-41C idle. Even @ default clocks on a default H/F.)
 
Huh, that's a shame. The Computronix guide on Tom's is gone, or at least the link he posted is no longer active. However, someone on Overclock.net has created a similar thread, referencing Comp's original guide.

If my intention was to pursue a massive OC, I'd probably think about calibrating and keep tabs on "the distance to TJMax", but as my plans are relatively tame, I doubt I'll bother. Thanks for the link though. Very interesting info.

Good to know it's relatively common (stupid Intel sensors) and that it doesn't appear to impact an overclock. I'll still reseat the sink as I did not follow the AS5 guide to proper thermal paste application for quads (didn't realize it was different from my previous C2D). Maybe I'll shave off a few degrees, but I'm not holding my breath for consistent temps across all four cores.
 
Mines does that too. Core 1 is the highest and core 2 is the lowest
tempk.jpg
 
Last edited:
I have: 35, 36, 40, 36 on the 4 cores. I don't know...seems normal to me. Never had any stability issues.
 
Huh. Interesting stuff. Thanks for all the personal comparisons.

Now, I realize lapping isn't necessary, but if I were to consider it (I understand all the potential downsides including voided warranty and ultimate destruction of the processor should I screw up) what should I use to protect the pins? The Q9550 does not come with one of those hard black plastic rear guards, so I'd need to come up with something. One suggestion I've heard is anti-static foam. Any other possibilities?

... mind you, I haven't committed to lapping. Simply considering it.
 
I'm not sure lapping is going to help you much. If it would help you, I would think you'd have higher temps across all the cores. If there's an issue with the heatspreader, the only way you'd be able to fix it is to take it off.
 
Is there a chance the issue is caused by Core #1 being located in a "dip", a shallow area in the heatspreader where the sink isn't making good contact? Wouldn't that be something lapping could potentially fix?

Bascially, it's a "what if" scenario for me. If I don't lap, I'll be left wondering, "What if I had?"
 
Include me in this happy club of confusion

Core 4 is almost 9 degrees cooler than the other three.
 
I wish I had that problem. It's definitely easier to explain away as a bad sensor. With one core being hotter than baseline, there's always the possibility of it being fixed with a heatsink reseat, reapplication of thermal paste, lapping, etc. When one core is 10 degrees COOLER ...? Especially if you have 33C idle and one core rockin' frosty at 24C like yujie. I guess it does make you wonder why all four cores aren't 24C ...

And 39C max, yujie? Damn. Either you're water cooled or I have some seriously piss poor airflow (or maybe my AS5 job really was just that bad).
 
I honestly don't know why you're so concerned about this. Regardless of the cause, it is really not a significant thing. Uneven temps aren't bad for your CPU, nor will they limit your overclocking potential.
 
i have had many, many duallies and quads, more than half have been e0 stepping. and i always get those type readings. the fact that its is 8 or 10 degrees higher on load and idle makes me think that particular core is not contacting the heatspreader as good as the others. in which case no amount of lapping would help. the other option is that it could be the normally flaky intel heat sensors on 45nm c2d.
 
I'm leaning away from lapping. If I were reaching for 4GHz or something, I'd likely do it. But 3.4-3.6? I know my rig is more than capable of 400-430 FSB no problem w/ minimal heat issues. Still, I'll be reapplying the thermal paste in the near future. I'll report back with results.
 
You must log in or register to reply here.
Back
Top