uneven cores?

[zigzag03]

I don't know if this is a cpu question or a cooling question, so i'll post in both places, with apologies for the redundancy...

My recently acquired Q9300 (really living on the cutting edge here as you can see) shows uneven core temps, or more specifically, one core that is much cooler than the other three, 30c vs 45c @ idle, similar when pushed. I've looked at the paste coverage, looks good to me. The liquid cooler is new and untested (and I could write a book about that) but have no reason to think it's the problem. Is it unusual for this processor to behave this way? My previous unit was a Q6600 and it was very consistent across the cores.

Grateful as always for the input of those more experienced, zz03

 

[ShintaiDK]

I wouldnt worry. There can be many reasons on why. The concave of the IHS, or that one core on the chip is simply a super bin. While the neighbour is not.

 

[tweakboy]

This is normal,,,,,,, its concave as shintadk said This is why people delid or lap CPU.

This is why you use thermal paste for. gl

 

[fuzzymath10]

Yep, I have one core on a Q8200 that never registers below 43C. At load, all four cores even up, so I don't really care too much. It's probably a meaningless comparison, but for whatever reason I don't see this on both my 1155 CPUs. I don't worry either way.

 

[Techhog]

Yep, I have one core on a Q8200 that never registers below 43C. At load, all four cores even up, so I don't really care too much. It's probably a meaningless comparison, but for whatever reason I don't see this on both my 1155 CPUs. I don't worry either way.

Then you're fine. Idle is weird.

 

[alyarb]

every MCM chip i've ever worked with had a good disparity in idle temps. If the machine is on and idling, then you still have at least one core active just by running the system. It may not be very busy, but it's busier than the cores on the other die. Think of the architecture you are dealing with. You have a large circa 2005 bus that is your only way of getting to the rest of the system. So it's always on, there is no "deep idle" for the FSB. That bus is being split into a Y, and half of it is going to the active core and the other half going to the less active core. That creates disparity too. And nobody said anything about your two dies being of the same quality from the same wafer. I have seen Q series demote themselves to E series before.

On modern chips, they are much more SoC-like than simply dedicated execution cores like Yorkfield. Doesn't matter if they are dual core or 12-core, the only external buses you are going to see nowadays are basically PCIe and DDR3. you have intelligent load balancing in the system agent and most of the rudimentary system traffic occurs on a ring-shaped bus going around a single monolithic die. Gigantic shared caches have begun to really hog the relative area on the device too. All of this removes any disparity in temperature among the different die regions or wherever your Nth core temp diode is located.

 

[VirtualLarry]

I would say, unless it's overheating, don't worry about it. Some disparity at idle is normal.

 

[zigzag03]

thank you all for your responses. i have only just put this old thing back together, and now i'll try to crank up the action and see what happens. meanwhile, i won't sweat it at idle!