• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Normal Core temperature Discrepancy?

S

slavearm

Limp Gawd
Joined
Apr 7, 2005
Messages
179
I am running a core i7 as noted in my sig. I recently noticed that I have a pretty significant temperature difference (just this morning actually) between core 0/1 and 2/3. 0/1 are at 57C while 2/3 are at 40C. Maybe I should remount my water block eh?
 
Load temp at 100% or idle?

At idle your cores do take care of different applications within windows. Core 0/1 may be (for me are) busier than 2/3 most of the time.

But if at load, a temp variance that big might point to a mounting issue.
 
What are you using to monitor your temperatures? And are these idle or load temperatures? If idle, try loading your CPU with Prime95 or IntelBurnTest and see if your temps are closer together.
 
The Core i7 sensors tend to be very consistent. A reported 5C difference is normal between core 0 and core 3 at full load when running Prime95 Small FFTs. Core 3 almost always reports 5C lower during this test. Anything beyond that is typically a problem and remounting your water block would be a good place to start.
 
That was just sitting around chewing on pudd. I was monitoring with everest... I am going to crank up a little prime session right now and heat up my office to see how much difference there is.
 
Actually... there it is. After running prime for about 5 minutes, the difference is between 5-7C... so I guess everything is normal then.
 
Be careful with Everest. The last time I checked, it was only reading temperature data from the first 4 threads of Core i7 CPUs when hyper threading was enabled. The first 4 threads only corresponds to core 0 and core 1 so the temperature data that it reports for core 2 and core 3 was actually coming from the first two cores.

Perhaps that bug has been fixed in recent versions but I'm not sure. Core Temp and RealTemp don't have this problem. RealTemp goes one step further than Core Temp and organizes the data so it is always coming from the correct physical core. If the APIC ID reported by RealTemp is not 01234567 then other programs may report core temperature data from the wrong cores. This was a bigger problem with the Core 2 Quads.
 
TBH, I do not have an i7. In my reading, however, it seems that cores are utilized when needed, and can push past the rated speed when only using one. This could lead to greater temps, I would suspect. (2.66ghz quad, using only one core, goes to 2.83ghz for a single threaded app. making the core hotter).

Someone please correct me if I am wrong.
 
Mr34727: What CPU do you have? Post a screen shot of your details if you can. You can upload to imageshack.us and then just post a link here.
 
Under full load, up to 6-8C discrepancy is to be expected simply due to the nature of the positions of the 4 cores. I don't know this for sure, but I'd be willing to bet that it's always the 2 cores in the center that are the hotter ones due to being surrounded on both sides by simultaneously hot cores. I get the same thing on my Q6600 with a lapped Tuniq Tower.

Now, a 17C difference is either a reporting error or you should definitely remount.
 
Can't remember how my load temps vary but my idle ones vary by up to 8C (usually 6C).
 
krameriffic said:
I'd be willing to bet that it's always the 2 cores in the center that are the hotter ones due to being surrounded on both sides by simultaneously hot cores.
Click to expand...
There are no cores in the center; they're arranged in a square pattern.
 
Never mind, I was wrong. The four dies are lined up, not arranged in a square pattern.
 
Here's some good pictures from the Intel press kits.

Core i7:
http://download.intel.com/pressroom/kits/corei7/images/Nehalem_Die_callout.jpg

Core 2 Quad:
http://www.intel.com/pressroom/kits/quadcore/images/2006.int.qua.txt.EN.13x18.jpg

krameriffic: In the Core 2 Quad design, the center two cores are not physically connected. These early Quads are basically two separate Dual Core dies sitting under the same IHS.

One theory I have is that a lot of temperature difference users see on these Quads is Intel using two different TJMax values for each Dual Core within the Quad. This is not documented anywhere but is the only reasonable explanation I've been able to come up with after looking at 1001 screen shots.

I think Intel might deliberately offset them by an average of 5C so all 4 cores don't reach the thermal throttling point at the exact same time when things get hot. This would make the reduction in power more gradual during a thermal throttling episode. Just an opinion.

Run a program like Prime 95 Small FFTs which equally loads all 4 cores on your Q6600 and if your temperatures get high enough, this 5C difference is usually obvious. This same 5C difference shows up time after time during this test when comparing core 0 to core 3 of Core i7 CPUs.
 
You must log in or register to reply here.
Back
Top