Sign in with
Sign up | Sign in
Your question

Temp Difference Between Cores?

Last response: in CPUs
Share
October 12, 2006 12:47:02 PM

Greetings.

I've got a new system with a C2D E6600 with a Zalman CNPS9500 AT cooler. When I mounted the cooler, I used ArticClean 1&2, followed by Acetone, and then Isopropyl Alcohol 91%. Then applied AS5 as directed by their website.

When idle, Intel TAT shows Core 0 at 31C, and Core 1 is at 34C. There is always a 2-3C temp difference between Core 0 and Core 1. CoreTemp shows the same results.

Under 100% load using TAT, Core 0 is reading 47C and Core 1 is 51C.

I'm not concerned with the overall running temps (I think they are pretty good for air cooling), but I'm wondering if this temperature difference between cores is normal, or whether they should both be running the same temperature? Is it possible that I got a "bad" chip?

Just wanted to get this all cleared up before it's too late to RMA any parts if necessary. Thanks for any input.

More about : temp difference cores

October 12, 2006 7:34:57 PM

I was reading the post on lapping the IHS on the CPU. Think that may help? I'd hate to void the warranty if that doesn't work.

Thanks.
October 12, 2006 8:14:42 PM

more than likely just a variance in the thermal diode in each core.. as much as they try, they can't always get every diode and transister identical in silicon wafers.

also, the chip that reads each diode may have a variance as well, these combined could stack up to quite a disparity.

I will use an analogy that might be easier to understand.. take 3.5 floppies.. they have a tolerance for the head position... if your drive is in the middle of the tolerance (perfect alignment), you can read/write most any floppy created on another drive.. however, if your to the + side of the tolerance, and the disk your reading was written by a drive on the - side of the tolerance.. good luck... you just added your tolerances up to an amount that wont work...

kinda similar to the cascading of the tolerances found in temp sensors and the circuits that read them
Related resources
a b à CPUs
October 12, 2006 8:25:06 PM

For what its worth, my cores always are about 2C off from one another with #2 being hotter. You're not alone.
October 12, 2006 8:34:56 PM

I'd be pretty happy with those temps.

I do recall reading a possiblity of the Zalman 9500 bottom may not be totally flat as well, and that lapping it produced better results. Although, that was before dual core came out.

Still, I'd say those are good enough temps.

Only question I have, is what method did you use to put the themal grease on (I know you said according to the website)? I'm guessing the thin line, in the center with the ( o ) to your top left, and the ( < ) to the lower left?
October 13, 2006 3:15:07 AM

Quote:

Only question I have, is what method did you use to put the themal grease on (I know you said according to the website)? I'm guessing the thin line, in the center with the ( o ) to your top left, and the ( < ) to the lower left?


Yes. I used the thin line of grease in the direction specified on the AS5 website. I carefully placed the HSF onto it, and gave it a slight twist in each direction (as the directions said) before tightening it down. Seemed to work pretty well.
October 13, 2006 3:45:08 AM

The same as different chips (of the same model) have variable overclocking potential, they may have slightly different heat output. Nothing to worry about, a friends build has about a 3C under load difference, we checked twice to make sure heatsink was seated well (Danger Den TDX) and it was. Just a variant. Nothing to worry about!
October 13, 2006 3:54:18 AM

My first core is always 3 C higher than my second load or no load. I am the guy that lapped my cpu today and it is still 3 C no matter what. But overall about 6 C less both cores after lapping.
!