EVGA

EVGA Z170 Stinger - Freezing Issues

Author
edirol
New Member
  • Total Posts : 10
  • Reward points : 0
  • Joined: 2016/07/03 20:49:48
  • Status: offline
  • Ribbons : 0
2016/07/03 23:23:47 (permalink)
Hey,
 
I have a EVGA Z170 Stinger and I was doing stability testing and for the longest time I was trying to figure out why my system was unstable with Corsair DDR4-3000 CL15 RAM when running Prime95 28.9.  I initially thought it was a RAM incompatibility, so I dropped my RAM down to 1600Mhz and it still froze when running Prime95.
 
Looks like this might just be a bug in Skylake CPUs and how they execute AVX instructions.
 
Prime95 versions newer than 26.6 utilize Intel AVX instructions which trigger the freeze.
So far when testing with v26.6 my system has been stable.
 
This bug can be fixed via a BIOS update, so can someone from EVGA confirm if v1.08 has indeed fixed this issue, or will you be releasing a new BIOS version to fix this problem?
 
Thanks,
- Edirol
post edited by edirol - 2017/02/22 18:11:59
#1

4 Replies Related Threads

    edirol
    New Member
    • Total Posts : 10
    • Reward points : 0
    • Joined: 2016/07/03 20:49:48
    • Status: offline
    • Ribbons : 0
    EVGA Z170 Stinger - Freezing Issues 2016/07/04 23:19:40 (permalink)
    OK, looks like I was wrong about Prime 95 v26.6 being stable on v1.08 BIOS.  System still crashed.
     
    New developments:
    BIOS v1.05 (what came with the mobo) appears to be stable with Prime 95 v28.9.  Still testing.
     
    Two Odd behaviours:
    1.  With XMP enabled, C-states set to auto will not allow the 6700k to clock up to 4.2Ghz via Turbo Boost.  I had to manually set C-states to enable so that it would boost to 4.2Ghz.  On another thread someone observed that if XMP is not enabled, it will turbo boost to 4.2Ghz.
    2.  EIST doesn't appear to be working correctly.  System will only idle to 1600Mhz instead of 800Mhz.  I haven't found a way to get it to clock down to 800Mhz.  Multiplier appears to only go down to 16x min.
    post edited by edirol - 2017/02/22 18:12:14
    #2
    edirol
    New Member
    • Total Posts : 10
    • Reward points : 0
    • Joined: 2016/07/03 20:49:48
    • Status: offline
    • Ribbons : 0
    Re: EVGA Z170 Stinger - Freezing Issues 2016/07/13 00:21:22 (permalink)
    So there appears to be a bug in v1.05 where if you are using the iGPU, it will only clock it to 300Mhz.  I had to drop down to v1.04 which doesn't have this bug and correctly clocks the iGPU to 1150Mhz on my 6700k.  Other than that both v1.04/1.05 was stable in Prime 95 v28.9 /w Blend test for 24h.
     
    Interestingly I did see the core clock down to 800Mhz with v1.04 though.  Not sure if that's a coincidence as only after my system was stable, did I start to use it normally with low load instead of stress testing.
     
    So that's it for me.  Hopefully these posts help someone.  It's unfortunate that 1.07+ introduced instability when stress testing with both Prime95 as well as Intel Burn Test.  My suspicion is that somehow when EVGA added microcode fixes for the Skylake Prime95 bug, they introduced other issues.

    Reference for Skylake Bug:
     
    Ciaos,
    - Edirol
    post edited by edirol - 2017/02/22 18:12:26
    #3
    edirol
    New Member
    • Total Posts : 10
    • Reward points : 0
    • Joined: 2016/07/03 20:49:48
    • Status: offline
    • Ribbons : 0
    Re: EVGA Z170 Stinger - Freezing Issues 2017/02/21 19:39:43 (permalink)
    Update:
    So every time a new BIOS is released for this board I test it to see if it still freezes Prime95 with default settings.  With the new 2.01 release it was no exception. The system still froze with BIOS defaults.  That got me wondering as all the Windows BSODs were the ones you'd typically see if you were overclocking without enough voltage.
     So with these settings my system appears to be stable /w BIOS 2.01
    - Vcore = Adaptive
    - Target Voltage = Auto
    - Offset Voltage = +50mV
     
    Now what doesn't make sense is if I run BIOS 1.04 the system is stable with default settings.  Not only that, I can set the Offset Voltage to -88mV and the system is stable.
     
    So it seems the behaviour of the Adaptive Voltage Control must have changed significantly after BIOS v1.05 and the problems I was experiencing probably wasn't related to the Prime95 Skylake bug.
     
    Hope this helps.
     
    Thanks,
    - Edirol
    post edited by edirol - 2017/02/22 18:11:07
    #4
    somethingc00l
    EGC Admin
    • Total Posts : 752
    • Reward points : 0
    • Joined: 2009/03/14 13:44:57
    • Status: offline
    • Ribbons : 11
    Re: EVGA Z170 Stinger (BIOS 1.08) /w 6700K Triggers Prime95 Skylake Bug 2017/02/22 15:40:24 (permalink)
    The Prime95 bug only occurs with AVX, by default Prime95 28.9 will use AVX2/FMA3. You have to create a local.txt file in the Prime95 folder with CpuSupportsFMA3=0 in the file to use AVX. So I doubt you are hitting the AVX bug, probably just plain old power delivery or heat problems. Source: http://www.pcworld.com/ar...r-the-skylake-bug.html

    #5
    Jump to:
  • Back to Mobile