Page 3 of 3 FirstFirst 1 2 3
Results 21 to 28 of 28
  1. #21
    ROG Guru: White Belt Array
    Join Date
    Jan 2012
    Reputation
    10
    Posts
    94

    Quote Originally Posted by Raja@ASUS View Post
    I'd wait until the memory arrives, as memory type/speed can impact how far you can overclock the system. On these CPUs, given temp constraints, I'd utilize the AVX offset parameter. We can deal with later, though.
    3 and 5 is the offset I'm using (before setting that all cores once hit 100/101 and she throttled down, again that was on the previous block) so yes I set those down.

    Ill wait for the ram, shes stable atm but I'd prefer to not have to go through this again and again (with my old x99 platform with the 59XX EE it was a mess, 69XX wasnt so bad).

  2. #22
    Tech Marketing Manager HQ Array Raja@ASUS's Avatar
    Join Date
    Apr 2011
    Reputation
    161
    Posts
    7,454

    Quote Originally Posted by Aysberg View Post
    When running multi threaded benchmarks, like the one in CPU-Z I can see an clear improvement from 42 to 44, but from 44 to 46 and 48 it stagnates, while single core gets better quite linear. From 44 to 46 also temperatures and wattage stays nearly the same. So I guess something is not working as intended.

    CPU-Z ST 497,9 / 522,6 / 546,4 / 571,0
    CPU-Z MT 10577 / 11426,9 / 11498,5 / 9593,8
    MT Ratio 21.25 / 21.87 / 21,04 / 16.80

    PS Maybe I need to do more testing, but somehow 46 and up feels not right (48 is quite obvious throttling seeing the MT result), but 48 is not my primary goal,at least not for 24/7 use.

    And thanks for your help.

    You can debug whether it's VCCIN related throttling by increasing the voltage to that rail. Set it to 2.1 to 2.2V and see if you get the same results.

  3. #23
    ROG Guru: White Belt Array
    Join Date
    Jan 2012
    Reputation
    10
    Posts
    94

    OK done, new mono block, cleaned the entire system out etc, fairly large different in temps, at stock 55 with all CPUS maxed out

    Overclocking time, AVC off sets 3 and 5, increased vcore to 1.15 and she just power off and starts back up again blah (vrm temps are WAY down, kinda nice)

  4. #24
    Administrator Array Silent Scone@ASUS's Avatar
    Join Date
    Mar 2016
    Reputation
    124
    Posts
    1,823

    1.15v at what frequency?

  5. #25
    ROG Guru: White Belt Array
    Join Date
    Jan 2012
    Reputation
    10
    Posts
    94

    Quote Originally Posted by Silent Scone View Post
    1.15v at what frequency?
    was testing 46 testing 44 now

    Again just a straight power off and starts back up

    Temps show in the 70s when this happens so cant be over heating

  6. #26
    ROG Guru: White Belt Array
    Join Date
    Jan 2012
    Reputation
    10
    Posts
    94

    on auto the voltage for 44 and 46 are in the 1.2X range but she completes all benchies without crapping out, but temps are in the 90s

    Ideas?

  7. #27
    ROG Member Array
    Join Date
    Apr 2017
    Reputation
    10
    Posts
    16

    Hey guys,

    Just wanted to share some of my findings with you for my 7980XE (non-delided and cooled with H115i) and R6E motherboard BIOS 1102.

    I've been stress testing the 7980XE CPU for a couple of weeks now and was able to find some good Core voltages for several frequencies.

    * NON-AVX Stress testing:

    These non-AVX voltages have been found with the following BIOS settings:

    CPU Load-line Calibration [Auto]
    CPU Current Capability [140%]
    CPU Power Phase Control [Auto]


    My non-AVX stress testing methodology involves the following:

    - Cinebench 20x "Run" in a row
    - Running Prime95 v26.6 (non-avx version) blend test 24h
    - Running Prime95 v26.6 (non-avx version) 8K FFT test 20 min (to make sure temperatures stay below 94C)

    Maximum stable Vcore offset mode:
    x42: -0,095V and above
    x43: -0,088V and above
    x44: -0,084V and above
    x45: -0,040V and above


    Once I found the above minimum non-AVX stable Vcore I realised I didn't stress test AVX offsets... without touching the LLC I wasn't able to get any AVX-512 offsets stable... I tried AVX-512 all the way down to an offset of 20. So I decided to increase the voltage. AVX offset of 4 was stable.

    I ended up with a non-stable config for AVX-512 ... so I had to stress test AVX-512 and tweak the voltages a little.

    AVX offset: 4
    AVX-512 offset: None stable


    * AVX/AVX-512 Stress testing:

    These AVX voltages have been found with the following BIOS settings:

    CPU Load-line Calibration [Auto]
    CPU Current Capability [140%]
    CPU Power Phase Control [Auto]


    My AVX stress testing methodology involves the following:

    - Running Prime95 recent version (with AVX) blend test custom 1h for AVX
    - Running y-cruncher HNT stress testing for more than 100 passes to test AVX-512
    - Running Prime95 v26.6 (non-avx version) 8K FFT test 20 min (to make sure temperatures stay below 94C, because we are increasing the VCore this time again)

    Maximum stable Vcore offset mode when stress testing AVX-512:
    x42: -0,065V and above
    x43: -0,045V and above
    x44: -0,025V and above
    x45: -0,020V and above


    AVX offset: 4
    AVX-512 offset: 8 (too much voltage required for anything below this value with LLC set to Auto)


    However, with the above settings I'm still getting some random freezes after ~51 passes of y-cruncher with HNT stress test... So I decided to increase the LLC to a value of 6, which seems to have resolved the issue.

    I'm currently trying to tweak the voltages with the new LLC value set to 6 to get AVX-512 stable. So far this is my BIOS configuration:

    CPU Load-line Calibration [Level 6]
    CPU Current Capability [140%]
    CPU Power Phase Control [Auto]


    * Regarding Core temps:

    This is the list of worst cores to best cores (values taken from Sync All 4.3Ghz, offset mode -0,050V, Prime95 26.6 8K FFT, Corsair H115i):

    #3 - 95C
    #13 - 92C
    #16 - 90C
    #4 - 89C
    #15 - 88C
    #7 - 87C
    #6 - 86C
    #14 - 85C
    #17 - 85C
    #11 - 85C
    #10 - 85C
    #2 - 82C
    #8 - 82C
    #12 - 82C
    #18 - 81C
    #5 - 80C
    #9 - 78C --> On my build, that core isn't stable at 45x and requires more voltage than others at 44x, setting a Vcore offset doesn't seem to work properly for it
    #1 - 78C


    I was wondering if you guys have the same core temp values.

    Something weird in R6E BIOS, Core #1 and #8 are marked as the best performing cores, but the temps show that #1 and #9 are actually the best cores.

    The y-cruncher HNT stress test for AVX-512 has been the trickiest to pass for stability so far. I've had freezes when setting cores at 44x/45x, which is why I decided to increase LLC to 6 (however, I have to redo the voltage calculation now).

    Anyone trying to achieve extreme stability or are you guys running softer stability tests?
    Last edited by Thireus; 01-23-2018 at 02:36 PM.

  8. #28
    ROG Member Array
    Join Date
    Apr 2017
    Reputation
    10
    Posts
    16

    Anyone? At least to share the temps at max load for all cores, that'd be very helpful.

Page 3 of 3 FirstFirst 1 2 3

Posting Permissions

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