Temp Variation Between 4 Cores on i7-930 Help!

Discussion in 'Overclocking & Cooling' started by sohosources, Apr 2, 2010.

  1. sohosources

    sohosources n00b

    Messages:
    56
    Joined:
    Oct 7, 2006
    Hi, Gang,

    I'm starting to play around with my i7-930 / ASUS Rampage II Gene build...and I'm noticing quite a temp difference on core 0 -- it's about 8 C cooler than cores 1, 2 and 3 under load (and about 4-5 C cooler at idle).

    I'm not going for a killer overclock, as this is my main productivity workstation. I've OCd lots of other procs, mostly AMD, over the years...but this is my first i7.

    When running at 3.5 GHz while running a LinX burn test, Realtemp and CoreTemp both report temps of 60 C for core 0 and 66-69 C for the other three cores. Prime 95 runs about 2 C cooler across the board.

    I'm using a very solid Silverstone 850-W PS and a Hyper 212+ cooler. I'm not thrilled with the CPU contact patch on the Hyper 212+, and I've reapplied my AS Ceramique twice, seeing NO CHANGES in temps.

    Is the temp spread potentially a problem?

    Is a 70C LinX test temp too high? (I know it'll never get anywhere near that in daily, non-gaming usage, right?)

    Thanks,

    --KK in MN
     
  2. Psi*

    Psi* [H]Lite

    Messages:
    99
    Joined:
    Dec 24, 2009
    I do not have this CPU ... but what I do have ought to be close enough ... in short, worry not. what you have might not be the most typically reported, but it is not unusual for any version of i7s.

    I have a 10 C variation across mine. Take a look at this i7 980x ... it has a 12 C differential
     
  3. SicKlown42012

    SicKlown42012 2[H]4U

    Messages:
    3,319
    Joined:
    Jul 6, 2008
    SInce Intel has a temp sensor in each individual core, and since each core gets a different load put on it by the OS's scheduler, variation is completely normal. Also, since the cores in a different physical space, they each have a different amount of contact with the heat-spreader and then cooler.
    I've yet to see an Intel quad-core that didn't have less than 4-5c difference across the cores.
     
  4. sfsilicon

    sfsilicon Limp Gawd

    Messages:
    505
    Joined:
    Nov 1, 2005
    Does the Hyper 212+ have push pins? It's hard to ensure that you mounted the heatsink 100% flat with push pins. It helps to have someone hold it flat while you press the pushpins in.

    My Cogage TS uses the newer bolt retention. On my first mount I had 4-5c difference between 2 sets of cores. After redoing TIM (less is better) and then slowly doing an even turn by turn tightening of the screws I now have 3 cores at the same temp and 1 at 1c higher.

    The orientation of your heatsink might also impact your temps. I aligned the heatpipes so that they were parallel to the long side of the die.

    Finally you might have a slightly uneven CPU heat spreader and/or heatsink surface. Only way to fix that is lapping. It's quite a bit of work, so I stay away from it. I wouldn't do the CPU if it still has warranty.
     
  5. sohosources

    sohosources n00b

    Messages:
    56
    Joined:
    Oct 7, 2006
    My 212 Plus doesn't have the goofy pushpin mounts. Personally, I really dislike those!

    From what I've been reading / hearing, the variation isn't a big deal. Glad for that!

    --KK in MN
     
  6. liquidfluidity

    liquidfluidity 2[H]4U

    Messages:
    2,878
    Joined:
    Dec 28, 2006
    Variation isn't a big deal although , imho , that's quite a variance. I might try to reposition the hsf and make sure the screws in the mount are tightened evenly. My variance (920) is .5-1*C. Maybe I'm just lucky. About your temps - i7s are generally good up to 85*C loaded.
     
  7. MrGuvernment

    MrGuvernment [H]ard as it Gets

    Messages:
    19,169
    Joined:
    Aug 3, 2004
    i have heard people say above 10c difference could be a bad seat, anything below that is normal.
     
  8. sohosources

    sohosources n00b

    Messages:
    56
    Joined:
    Oct 7, 2006
    After carefully reseating and "re-TIMing" my cooler three times...my temps are exactly the same...

    I have three cores that all stay with 1 C of each other...and one "cool core" that runs 7-8 C less than the others.

    I have decided to run this box at 3.8GHz. In Prime95 and LinX, my idle temps are 38C and full load, with all 8 threads maxed, is 68C (Prime) 70C (LinX).

    Absolutely no trouble with this nice mid-range OC. 1.2 V Vcore, 1.6 V RAM.

    From what I've read, anyway, I don't think my cool core will be a problem. Hope not, anyway!

    Thanks,

    --KK in MN