Announcement

Collapse
No announcement yet.

Asux P9X79 + 3930K stable CPU, unstable under heavy PCIe load

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Asux P9X79 + 3930K stable CPU, unstable under heavy PCIe load

    Hello guys,

    This is a vexing problem I'm a bit on a crusade against at the moment since I'm completely lost as to what to do.

    Basically, the situation is as follows:

    * System is completely stable under any CPU/RAM-load I can throw at it - linpack/intelburn, prime95, video encoding, you name it
    * System is completely stable with Furmark even with GPU's factory OC

    However: System crashes to black-screen/hardlock/hardreset/spinning GPU fans as soon as I put serious load on the PCI bus. This mostly means 3D games. I can insta-crash my box with running DCS World (advanced flight sim), Heaven benchmark lasts about 45 seconds.

    Here's the kicker: My GPU is factory-OC-ed by about +80. If I underclock my GPU by -100Mhz, and set it's "Power Target" to 60%, most games run good enough for playing. It's not STABLE though, it just mitigates. DCS still crashes almost immediately, and I saw Heaven BM conk out after 30 minutes.

    All crashes are a black screen with the GPU fans SOMETIMES spinning up. There's no error logging whatsoever, no BSOD log, nothing in Event Viewer - feels like a power issue/fuse trip to me.

    Now, I've contacted Asus support (and poked the NVIDIA hornets nest as well), and they asked for a bunch of details and forwarded my problem to their backend tech support, but I'd rather not sit around on my hands doing nothing. :) I'm sure there's a fix somewhere ..

    Has anyone any idea what I can try to fix this? Someone pointed me at this forum saying you are the true experts. (cheap shot, I know, apologies)

    In any case, thank you very much for reading!

    ---

    Here's my system, and what I have already tried to fix this:

    - gpu: asus 780 ti dc2 oc (temps are fine, 30 idle, ~70 load)
    - intel i7 3930k (rev C2 so it does gen3), no overclocking, temps fine (32 idle, 60 6-core load), corsair h100
    - asus p9x79 pro bios 4608 (latest)
    - ram: Corsair CMZ8GX3M2A1600C9 x 6 = 24GB
    - psu: corsair ax760 (new, works) - old: hx850, replaced on suspicion
    - dell up3214q 4k@60hz over dp/mst (tested also: 1080p screen over hdmi)
    - samsung 840 pro ssd
    - win7 64bit, win8.1 64bit (clean install on other disk)
    - had a 7970 DC2T, which ran more-stable-than-the-gtx so I didn't bother investigating the bimonthly crash which I now suspect to have the same roots. Don't have it anymore though. The 7970 has significantly lower power draw under full load, fwiw.

    things tested that DIDN'T help:

    - swapped GPU with TWO new identical ones
    - swapped PSU (old: corsair HX850, new: AX760)
    - swapped/tested RAM each individually, memtest86+
    - updating GPU bios from 80.80.34.00.A01 to A02
    - bios reset to defaults
    - tested drivers: 331.82 (from asus), 332.21, 335.23, 337.61
    - fresh windows 8.1 install on different disk with newest NVIDIA drivers only, no crapware, no addons, no unneccessary devices connected, nothing
    - no nvidia 3d/physx/experience installed
    - swapped display and display cable (type), tested different 1080p screen
    - swapped PCIe slots (tried all)
    - used nvidia-pcie-enabler to allow driver use pcie gen3
    - setting PCIE link speed to gen2 and gen1 in bios
    - disable cpu C-states, disable cpu speedstep, disable intel turbo
    - run DRAM at 1066Mhz, 1333Mhz. Verified CAS latencies and voltage. Ran w/ XMP profile.
    - raising GPU Minimum mV by very small amount (0.950 → 1.0), did not touch max voltage
    - tested older mainboard bios revisions
    - raising VCSSA voltage to 1.1V
    - raising base block to 103Mhz
Working...
X