80C *too* hot for Sandy Bridge?

dderidex

Supreme [H]ardness
Joined
Oct 31, 2001
Messages
6,328
Using a Corsair H70, which I guess I'm not TOO over-excited about.

Anyway, running an I7-2600K at 4.4ghz on all four cores, about 1.4v to the CPU*. Currently, my hottest core is 77C (Folding @ Home maxing it out all day), but Core Temp reports that it's peaked at 80C. Must have just been for a moment, as I didn't see it. Coolest core is 72C.

I'm running an Antec 300 case, with the three-speed fans (one is on the radiator) all set to 'medium'. If I set one of the front (intake) case fans to 'high' and the rear (exhaust, over the radiator) fan to 'high', it knocks off about 4C from the CPU temps, but I don't really like the volume increase. It's not AWFUL, but...I'd rather not.

Ergo, the question. Running in upper-70s with a peak at 80C is...good? Bad? Indifferent?

* The voltage varies. I'm using the BIOS 'offset' mode of "+0.100", which appears to be what my chip needs for 4.0ghz+. Which is VERY weird to me, as it's tested out (for days, using Folding at Home, Prime95, various games, etc) rock solid at 4.4ghz with that "+0.100" setting. If I bump it down to even "+0.095", then the CPU becomes unstable in several games (system hard locks), even down to 4.2ghz (I didn't try pushing it lower to see what clock speed it re-stabilizes at). Totally bizarre to me.
 
Anything over 72C is not good in my books, if it is just while doing stress tests that is ok because it will never get that hot during normal use. If it is sitting at that temp folding 24/7 then ya you will soon be in trouble.
 
I've had my 2500k up to 82-87c for hours at a time while transcoding. No issues thusfar, but I too am looking to cool it down more. I don't think a peak of 80c is going to hurt it, since TJm has been reported to be 90-100c.
 
Sandy Bridge can run hot. My 2600k for instance, at full loads (Prime95) when OC'd to 4.6Ghz, mine was running at 101C on one of the cores. But out of the Box at full load it ran about 84C. The specified throttle limit from Intel I believe is 100C. The Maximum safe operating temp they suggest is 95C before it starts causing damage. So anything below 95C should be fine.
 
In theory, it should run a 1C below TJmax 24/7 for years without any problems (although I don't know that I'd want to test that out). I think mid-70s are fine - you might consider trying to change fans on the H70 or something to bring that down a little, but I doubt you'll have any serious problems from running that temp 24/7.

According to Asus (and I tried it and it worked), you can reduce VCCSA, PCH PLL and PCH Voltage to get lower temps without affecting stability. I dropped them all 1 notch (I think it was) and it dropped my load temps 3-5C.
 
Keep it under 85C and you will be fine...You aren't going to hurt this thing unless you start throwing crazy voltage at it.
 
According to Asus (and I tried it and it worked), you can reduce VCCSA, PCH PLL and PCH Voltage to get lower temps without affecting stability. I dropped them all 1 notch (I think it was) and it dropped my load temps 3-5C.

Might give that a try. I'm using an Asus Z68 board, so definitely do have those options (the amount of tinkering possible in the BIOS is...amazing).
 
Might have been testing too many things at once, last time?

Took CPU voltage down to "+0.090" and load line calibration down to 'medium' (25%) from 'high' (50%). That seems to have knocked 6C off the temps.
 
For most people in here, they'll probably tell you it's a bit on the high side. As far as safety of the chip goes though, it's perfectly fine. I was running the stock cooler for a little while on my 2500k. Even at stock speeds, it topped out in the high 70's. If Intel is ok with those kinds of temps for their stock cooler, it's unlikely to be any danger to the chip.

I'm fairly certain these chips can actually run into the 90's before you're pushing danger, not that i'd recommend it.
 
You should be OK but keep an eye on it during long high cpu load sessions. My 2600K reaches 72 degrees with linx but never goes over 62 during prolonged video encoding and other similar tasks.
 
For most people in here, they'll probably tell you it's a bit on the high side. As far as safety of the chip goes though, it's perfectly fine. I was running the stock cooler for a little while on my 2500k. Even at stock speeds, it topped out in the high 70's. If Intel is ok with those kinds of temps for their stock cooler, it's unlikely to be any danger to the chip.


Unless I'm mistaken that's only true under "normal" voltages. Once you start throwing more voltage into the chip temperatures become a lot more important.
 
Totally fine. I have been running Intel CPUs at temperatures above 80C 24/7 for years without any issues.
 
"Heat will never kill our processors."
-An Intel Engineer

Sounds like a challenge! I think more and more people will want to run passive setups now that we're getting significantly better performance per watt (starting with SB, and hopefully advancing again with 22nm IB). - I know I want that in a HTPC / media/data server.

I wonder what would happen if someone just ran a passive setup right to the temperature shutdown threshold and kept it there... at the highest permissible "safe" voltage?

That test would be a waste of resources, but it's probably in line with Intel's envisioned worst case scenario.
 
Back
Top