+ Reply to Thread
Results 1 to 8 of 8

Thread: Possible Core Temp Diode bug

Hybrid View

  1. #1
    ROG Enthusiast Claviger +10
    Join Date
    Apr 2012
    Posts
    66

    Possible Core Temp Diode bug

    I tried two different BIOS setups last night.

    i7 3820, Rampage 4 Gene, BIOS 1305, 125x39 1.504 Vcore loaded, Prime95 with AVX Win7 SP1

    Turbo Mode: All Core 125x39
    Turbo Mode: Per Core 125x39 (39 set for all entries in the per core section to force 39)

    With All core mode temps under load were
    Core 0:71
    Core 1:67
    Core 2:67
    Core 3:70

    With Per Core mode set and no other changes temps were
    Core 0:77
    Core 1:77
    Core 2:76
    Core 3:77

    Running the exact same setups for all other settings, exact same software environment, and exact same thermal loading. Testing done within 5 minutes of each other in the same location with same ambient temps.

    Clearly there is some witchery going on with this and now I am not sure which setting's temperatures to trust, can anyone else duplicate this?

  2. #2
    ROG Guru: बोधि Arne Saknussemm +150 Arne Saknussemm +150 Arne Saknussemm's Avatar
    Join Date
    Apr 2012
    Location
    In the gutter but looking at the stars!
    Posts
    3,229

    That is an interesting result. I like the experiment! Maybe by core is more demanding than all core and needs more juice?

    When I get back to my PC I'm going to investigate this with my 3930k. I'll post back.....

    Phobya Bench Table / Rampage IV Extreme / i7-3930K @4.6GHz offset -0.015 / 16GB Corsair Dominator Platinum 2133 CMD16GX3M4A2133C9
    ASUS GTX Titan (Second Titan on the way) / ASUS Xonar Essence STX / 2x OCZ Vertex 3 256GB 2x WD Caviar Black 500GB
    Custom loop: EK Supreme HF Full Ni / EK-FC TITAN XXL Edition / 2x EK Coolstream XTX 360 / EK D5 X-Top CSQ Acetal
    Alphacool D5 Vario / EK-Bay Spin Res
    Dell U2711




  3. #3
    ROG Enthusiast Claviger +10
    Join Date
    Apr 2012
    Posts
    66

    I eagerly await your results, if it can be duplicated, then ill do some further testing. If not I'll pretend it didn't happen and move on lol. In Benchmarks I pull the same scores either way, infact, "All Core" scored the best for me by a .016 millisecond time in Wprime at the same speeds.

  4. #4
    ROG Guru: बोधि Arne Saknussemm +150 Arne Saknussemm +150 Arne Saknussemm's Avatar
    Join Date
    Apr 2012
    Location
    In the gutter but looking at the stars!
    Posts
    3,229

    OK, so I just went into my bios and changed to per core and set 46 for each. This is the only change I made. Normally I have 46 all core. I booted up.... and windows hung on the loading screen. Reset.... booted again BSOD 0x101.....so not enough juice.

    My overclock is rock solid stable for everything I have thrown at it! But I know I have shaved power/voltage down to the wire to run it. If per core does not boot for lack of volts it means per core needs more than all core even if the multipliers are the same.

    More volts higher temp which is what you were showing.

    I suspect your all core voltages could be shaved a bit if your per core boots just at a higher temp.
    Last edited by Arne Saknussemm; 05-21-2012 at 05:33 PM.

    Phobya Bench Table / Rampage IV Extreme / i7-3930K @4.6GHz offset -0.015 / 16GB Corsair Dominator Platinum 2133 CMD16GX3M4A2133C9
    ASUS GTX Titan (Second Titan on the way) / ASUS Xonar Essence STX / 2x OCZ Vertex 3 256GB 2x WD Caviar Black 500GB
    Custom loop: EK Supreme HF Full Ni / EK-FC TITAN XXL Edition / 2x EK Coolstream XTX 360 / EK D5 X-Top CSQ Acetal
    Alphacool D5 Vario / EK-Bay Spin Res
    Dell U2711




  5. #5
    ROG Enthusiast Claviger +10
    Join Date
    Apr 2012
    Posts
    66

    Good point, I'll take a look, would account for higher temps, but doesn't account for cores 1 and 2 suddenly being within 1c of 0 and 3 when on all core they are usually 3-6c apart.

  6. #6
    ROG Guru: Platinum Belt Zka17 +100 Zka17 +100 Zka17's Avatar
    Join Date
    Mar 2012
    Location
    Mount Pleasant, SC, USA
    Posts
    5,325

    Quote Originally Posted by Claviger View Post
    I tried two different BIOS setups last night.

    i7 3820, Rampage 4 Gene, BIOS 1305, 125x39 1.504 Vcore loaded, Prime95 with AVX Win7 SP1

    Turbo Mode: All Core 125x39
    Turbo Mode: Per Core 125x39 (39 set for all entries in the per core section to force 39)

    With All core mode temps under load were
    Core 0:71
    Core 1:67
    Core 2:67
    Core 3:70

    With Per Core mode set and no other changes temps were
    Core 0:77
    Core 1:77
    Core 2:76
    Core 3:77

    Running the exact same setups for all other settings, exact same software environment, and exact same thermal loading. Testing done within 5 minutes of each other in the same location with same ambient temps.

    Clearly there is some witchery going on with this and now I am not sure which setting's temperatures to trust, can anyone else duplicate this?
    It's interesting to see that with both setups the core temps are increasing from core#0 to core#4 in a pretty regular manner. Yes, it's known that there is a slight difference between core temps, but I never seen such a regular increase between them. It makes me wondering if the CPU cooler/waterblock is mounted properly... It seems that on one corner is more tightened then on the others... Or, maybe the thermal paste distribution?

  7. #7
    ROG Enthusiast Claviger +10
    Join Date
    Apr 2012
    Posts
    66

    It's been consistently disparate between the 4 cores through i think 6 mounts, with 3 types of TIM and two different water blocks. I can say that baring any odd physical IHS warping, my mount applies even pressure every single time since I use number of threads showing on the mounting studs for my water block to verify even torque. I have tried various pressure points, and while more pressure yields lower temps, there is still a 3-6c disparity.

    The disparity of core0/3 vs 1/2doesn't bother me, but the disparity of all core vs per core does :P

  8. #8
    ROG Guru: Platinum Belt Zka17 +100 Zka17 +100 Zka17's Avatar
    Join Date
    Mar 2012
    Location
    Mount Pleasant, SC, USA
    Posts
    5,325

    Yeah, but the surface of the CPU is not completely flat... But, yes, this was just a thought - if that doesn't bother you, it's fine!

+ Reply to Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts