Help with Core i5 2500k processor

[handsome_dave]

Hello

Sorry to start my first post with another 'temperature thread', but i'm worried that i'm seeing a huge temperature difference between cores. Core 1/3 run about 30-33C while core 0/2 run at about 40-43C. Are these temperature differences normal? Or should I RMA my cpu? This is my first build and I've put quite a bit of money into it so I would prefer things to work the way they should Btw, I am on stock frequency at 3.3Ghz and using a Corsair A50 cpu cooler. Room temperature is roughly 25-30C (its hot here in Toronto and I have no AC!). Thanks!

 

[Borealis7]

some basic things to check first: if you have spread the Thermal Grease on your CPU evenly, and if your heatsink is fastened properly.
what program are you using to monitor the temps?

 

[handsome_dave]

I am using Coretemp and HWmon (not at the same time, ofcourse). I am also using the Thermal paste that came with the Corsair A50 HSF. I applied 2 lines of thermal paste on the heat sink where the grooves are since it is a bit uneven here. Heatsink is screwed on the cpu pretty tight. I had the same issue with the stock heat sink fan; that is, core 1/3 were about 10-12C Cooler than core 0/2. I'm beginning to think my cpu might be defective:S

 

[handsome_dave]

I forgot to mention, that when I run prime 95, core temps are much closer together with maybe a 5-7C difference (Core 2 still the highest out of the bunch).

 

[MTDEW]

My 2600k does the same thing, core 1-2 are always 7c-8c higher than the two cooler cores 0-3

Ive reseated my heatsink about ten times, and tried every thermal paste method from the pea size center blob, two lines, and spreading it evenly, more compound/less compound and it doesnt make a difference, its always the same.

So i just learned to live with the fact of that is just how my cpu is.
Im guessing sometimes you get one that is just a bit more concave than others and that is just luck of the draw.

And im not going to bother going so far as lapping it trying to fix it as long as all my temps are good.

note: This is using a Noctua NH-D14 which just about impossible to install wrong due to how the mounting system works. (those who own one, know what i mean)

 

[MTDEW]

Here is what mine looks like under load. (prime 95 small FFTs)

 

[Diogenes2]

Normal .. Inner cores run up to 10c hotter on my 2600K ..

Do not go to trouble of re-installing HSF ..

 

[handsome_dave]

Thanks for the responses. So I guess its not defective but normal to have a 10C difference. I was only concerned because i've read people having nearly the same core temperatures, and I have a friend whose 2500k had core temps approximately 3-5C between them. Who knew it would depend on your luck to get a proper working cpu! :S

 

[Kenmitch]

My 2500K's water cooled. Idle temps are pretty much even on all cores. Under 100% load the 2 inner cores run about 6-8 degrees warmer. Never tried reseating my water block as it kinda seems to be the norm as far as I can tell. Running at 4.6ghz @ 1.36v(loaded) so my chip is not a great one

 

[handsome_dave]

My 2500K's water cooled. Idle temps are pretty much even on all cores. Under 100% load the 2 inner cores run about 6-8 degrees warmer. Never tried reseating my water block as it kinda seems to be the norm as far as I can tell. Running at 4.6ghz @ 1.36v(loaded) so my chip is not a great one

It looks like you might be the luckier ones! During idle, my core temps differ about 9-12 degrees:| Right now, its sitting at:

core 0: 38C
core 1: 33C
core 2: 42C
core 3: 30C

As you can see, I have nearly 12C temperature difference!!! under prime with small fft, i get about

core 0: 55C
core 1: 53C
core 2: 62C
core 3: 55C

Core 2 is my busted core

 

[betasub]

^ Or a busted sensor/calibration.

Stop worrying & love your in-spec temperatures. Leave the angst to those who by design or fate end up really pushing the official limits.

 

[handsome_dave]

^ Or a busted sensor/calibration.

Stop worrying & love your in-spec temperatures. Leave the angst to those who by design or fate end up really pushing the official limits.

You are right, but it took me a while to save up for what I have now. That's why I wanted to make sure everything is operating correctly 12C does seem pretty big to me and eventually i'll want to overclock but i'm holding back because of this temp difference. I do love my pc though, and I'm coming from a mac. Couldn't be happier.

 

[MTDEW]

Ooops...edited for quoting wrong post.

Back on topic: I guess my core temp differences aren't as rare as i thought.
Good to know.

 

[handsome_dave]

hehe, so I guess I shouldn't RMA my processor then? :S

 

[MTDEW]

hehe, so I guess I shouldn't RMA my processor then? :S

Nope, you should overclock it instead. :biggrin:

 

[BonzaiDuck]

Just some minor thoughts. I can't be sure for not seeing the phenomenon, and it's been two generations since the Wolfdale LGA-775's. Even the most recently release of the Wolfie's -- or among the recently released of spring 2010 -- was the E6700 LGA-775, and it, too, showed a "stuck" sensor at idle. but if lowering the room-ambient considerably shows a drop in the "high" core temperature, then that's not it.

And it's interesting that two of your cores are high, the others (about 10C?) lower.

But I'd been through this for that thread a week or so back -- "Are you lapping your Sandy Bridge?"

Yes -- you could say it's a waste of time, some sort of "risk," warranty-voiding-irresponsibility, whatever. To an equal degree, there is also the heatsink, which come with a convexity in the base -- now observed to be a design for at least two different manufacturers.

So when I first fired up my 2600K and got the OS and monitoring software installed, the room ambient was 78F, and all four cores were flipping back and forth from 25C to 26C. But idle temps don't mean much -- especially if there's a possibility that sensors are stuck. Mine just seemed to indicate that they were fairly well calibrated and all working.

Now I'm running PRIME-95 for just a prelim test for an over-clock setting -- the voltage is probably a tad too high because it needs further tweaking, or 1.36V under load with mild LLC. 4.4 Ghz . . . . room-ambient 79F this time, and load values of 64C, 70C, 64C, 64C for #0, #1, #2 and #3 respectively.

Frankly, I think they're a tad high, but then there's the loaded voltage. I can't account for #1, unless either the sensor is inaccurate, or it's just an extra-loaded core.

 

[handsome_dave]

Thanks for the reply BonzaiDuck, So your sensors must be well calibrated. My two sensors on core 0 and core 2 seem to be busted or maybe my heatsink is not put on properly. But even with the stock heat sink i was getting numbers that were 10C off. And the ambient temperature in my room is probably 35C today with the horrible heat so this is making my core 2 idle between 41-44. Core 0 idles at 48-41 and core 1/3 are sitting between 30-33 I'm still contemplating returning the cpu.

EDIT: Ooops, i meant 38-41, not 48-41, haha..

 

[BonzaiDuck]

Thanks for the reply BonzaiDuck, So your sensors must be well calibrated. My two sensors on core 0 and core 2 seem to be busted or maybe my heatsink is not put on properly. But even with the stock heat sink i was getting numbers that were 10C off. And the ambient temperature in my room is probably 35C today with the horrible heat so this is making my core 2 idle between 41-44. Core 0 idles at 48-41 and core 1/3 are sitting between 30-33 I'm still contemplating returning the cpu.

Where do you live? Here in So-Cal? I suppose AZ or Nevada could be worse today, but this is bad enough.

Is that Corsair A50 a "direct touch" with copper pipes exposed to the processor's IHS? or does it have your usual heatsink base -- possibly plated with nickel?

It could be your sensor -- which means . . . "live with it." Or -- get a reliably straight metal ruler, hold the HSF base up to a light and see if there are gaps, a convexity (like some are designed) -- if possible (and it's inconvenient, I know) do the same to the processor's IHS. Only you can make the decision about taking time to do this, since I think we'd concluded that your 2500K is just fine.

And I continue to recommend nano-diamond thermal paste -- IC-Diamond. That's worth a couple degrees C anyway. If you're going to reseat the HSF and clean off the old thermal paste, pick up a 7-carat tube of IC-Diamond and use it instead.

EDIT: Actually, let me back off on my "tone." OF course you should RMA the processor and replace it, if you think it's wrong. As I'd said several times in other threads, The more mature stepping of the Wolfdales and even the Yorkfield cores had this "stuck sensor" problem, and Intel issued a disclaimer to say they "weren't meant to accurately measure idle temperatures." But your situation seems different. If either the sensor is way off, or one core seems to be running too hot . . . if you think it's defective, then go for it.

Also, I'll add that I have one core that seems consistently 5 to 6C higher than the rest. OC'd, it gives me a tad bit of pause, knowing it's only a few C below the throttling threshold of 72+C under full loading. That's on a 2600K, OC'd to around 4.4 Ghz. this is preliminary -- I expect to make a lot of tweaks . . . so . . .

 

[handsome_dave]

I think i'm just going to have to take it up the A and just live with it Mainly because I bought this online from TigerDirect and to RMA it, i'd have to mail it back to the warehouse before they sound out another one. ugh! I'll try to re-seat the heat sink to see if that does anything but i'm doubtful. I live in Toronto, and temperatures here have been 27-30C and my room is tiny so when I turn on the computer, my room reaches up to 35C! ( I also have no AC). This could also explain why I have high idle temps.....