Intel Core i7-3820 Temperature Discrepancy Between Cores

mpitroff427

Honorable
Apr 12, 2013
20
0
10,510
I have recently built a new rig, and upon running core temp, I notice I have a bizarre temperature discrepancy between the four cores on idle. The first two (marked c0 and c1 in Core Temp), fluctuate between 29 and 31 degrees. However, the c2 core maintains a temp of around 37-40 degrees. The last core (c3) is a little better, staying at 35-36, sometimes jumping to 38 at odd intervals. I used this processor in another rig with the same heatsink, a Hyper 212 EVO, and didn't experience such strange temps. Should I try reseating the heatsink?
 
Solution
40 at idle is high. The discrepancy is normal though. My 3570k has a similar spread with a 212 evo. You should reseat it and make to sure to tighten the 4 screws evenly (alternating pattern) and completely.

Kjwinden

Honorable
Feb 9, 2013
415
0
10,860
40 at idle is NOT high, it could be the room you are in is hot or something like that (the CPU can never get below room temp regardless of how good your cooling is.

The multi-core processors are designed to not use all cores when not under load- so assuming you are not under full load when you are seeing these numbers what you are seeing is normal.

What you should really be worried about is what temps you see when you are under load. Idling at 40 but then seeing 62-70c is OK. However, if you idle at 30c and see temps of like 100-110c under load means there is a problem.

In closing, what you are seeing isnt problematic. What you really need to worry about is when you see high temps under load.
 
As wanderer stated the temperature differences are very normal to see on a processor. The reason why is due to the what is around the core and the ability for the heat to bleed off from it. To give you an idea on how this works C1 and C2 are on the edge and only dark silicon (not being used) to the other side of them, C3 is in the middle and the cache is above it and C4 is on the edge with only the cache above it. This means that there for C3 there is limited area for it to bleed off heat (the set up might be a little different but this gives you an idea on how it works).
 

mpitroff427

Honorable
Apr 12, 2013
20
0
10,510
Hm, well thanks for easing my fears! I was mainly worried though because I had a previous build that used the same CPU and heatsink (as well as motherboard) and I don't remember seeing such a difference between the temperatures. If I remember correctly, it idled around 33-34 on all four cores. I have added a pull fan to the hyper 212 EVO, so I'm not sure if that will cause any problems. Oh, I probably should've mentioned, my ambient temps fluctuate between 23 degrees 27 (no air conditioning).
 

mpitroff427

Honorable
Apr 12, 2013
20
0
10,510


I'm pretty sure I have it installed correctly (have the arrows pointing the outward from the heatsink for a push fan, the other fan was preinstalled as a pull). I won't be able to do stress testing until I get home in a few hours, but when I do, should I just look for higher temps on the problem core when it's under load? Also, if I decide to reseat the heatsink, what would be the preferred application method for the thermal paste since I'm using the hyper 212 EVO? I used a double line on the heat pipes before, but I'm feeling that directly applying the past to the CPU itself and spreading it evenly with a credit card might be the best method. Any opinions?

 

mpitroff427

Honorable
Apr 12, 2013
20
0
10,510
Alrighty, did a bit of stress testing using a large battle in Dragon Age: Origins (I know, not the best method), and the highest temp that was recorded was 51 degrees on the problematic cores, but all the others ranged between 46-49 degrees, so it seems to be normal. I'll run Prime95 and report back with my findings.

Update: So stress testing went well, only ran Prime95 for about 10 minutes since I'm not going to be overclocking anytime soon. Here were the highest temps that were recorded per core.
c0: 61°C
c1: 59°C
c2 (problematic core): 63°C
c3 (slightly problematic core): 62°C

These look pretty consistent and they remained stable during the test, with c2 hovering around 61°C for most of it. From the looks of it, I don't seem to have any major issues with the temps when the CPU is under load. Should I do anything else? Perhaps run Prime95 for longer?