OTOY Forums

Hi, Guys,

quick note on this topic. From what we managed to find with other guys on our Facebook group & other parts of web that this seems to be an issue with PLX equipped boards. Nvidia was contacted multiple times & is aware of this issue. SomeOne wrote on tomsHardware that nvidia was able to replicate this crash and is working on solving this.

For now, solutions would be:

* stick with older drivers & older OctaneRender version if You have some ongoing projects to finish,
* downgrade to w7 - seems some guys are able to get x99e ws working there without any issues with newer drivers & new build of OctaneREnder,
* try linux - again, for some it works,

in the end, hopefully solution is going to come from nvidia & these boards will be as stable as they were before.

Thank you Tom for the latest info and all the help you provide on these forums (both here and Facebook).

Hi i just built a salve system with asus X99-E-10G WS it has 3 video cards with netstor turbox attached total 7. When I send over a scene to render I get PC just rebooting with the crash or just completely freezing. is this basically the same problem you are having? I do have window 10 and I am afraid windows 7 wont even recognize 7 GPus Not sure what to do

There is also 6 pin connector that I never connected. Not sure what it is for. My evga 1600 does not even have this kind of cable.

i tried 382. drivers and i cannot even use them with because octane 3.08 says cuda driver version is too low.

This connector helps power for PCIe, I think that you should connect it to the power supply. Be carefull, chek the polarity. On my board, I've connected all power connector (Xeon overclocking and 7 GPUs)

itou31 wrote:This connector helps power for PCIe, I think that you should connect it to the power supply. Be carefull, chek the polarity. On my board, I've connected all power connector (Xeon overclocking and 7 GPUs)

can this also be the reason PC freezes as soon as octane starts rendering? And do i run a wire from VGA port or peripherals? Thank you

UPDATE: it works fine with drivers mentioned in the above post 382 and octane 3.06 so it is PLX bug. Returning the motherboard I guess who know when they will fix the bug.

This may not apply to the watchdog violation error of the original poster but may help some others:

My Watchdog DPC violation was triggered by high temperatures. The display GPU with 3 fans was blowing hot air at the rendering GPU installed below.

Updated / Edited - rephrased

What may have been the issue for my system may not be an issue with other builds and other variations of the Watchdog DPC violation.
- - -

I used one Asus GTX 1080 STRIX A8G for the display in the top PCI slot.
Below that the two Asus GTX 1080 Ti FE were sloted.

When I used whocrashed the error information told me that my specific Watchdog DPC violation error code could also be triggered by overheating.

http://www.resplendence.com/whocrashed


- - -

I switched the two Asus GTX 1080 Ti FE to the top two PCI express slots and assigned the Asus GTX 1080 STRIX A8G to the bottom one.
From that point on forward the watchdog DPC violation errors were gone for me.

My assumption was that the three fans of the Asus GTX 1080 STRIX A8G were blowing hot air directly at the Asus GTX 1080 Ti FE.
This seemed to have triggered the Watchdog DPC Violation after some GPU running time for me.
- - -

linvanchene wrote:This may not apply to the watchdog violation error of the original poster but may help some others:

My Watchdog DPC violation was triggered by high temperatures. The display GPU with 3 fans was blowing hot air at the rendering GPU installed below.

I've built multiple systems in the past (4-7 GPUs) & some of them were spitting this error under wrong drivers. Some of them have overkill radiators that keep all GPUs below 50C - this error has nothing to do with temperature if I would have to guess..

Yes, I can also confirm that this is not a heat issue. My cards are hybrid and hardly ever get hotter than 45 degrees even under full load. Also, the watchdog errors are not just during rendering, even scene loading could cause problems.

I think as others are saying PLX chip on Windows 10 with post Nvidia 382 drivers is the fatal combination. Hopefully Nvidia can fix that.

itou31 wrote: Have only problem when I've played with vbios. Now revert back to original vbios to pass the cuda 9.10 drivers (from 387.92).

Hello,
Could you elaborate more on this? I am having the same problem as the op. Going nuts myself here.
Thanks!

Hi FAZ,
I have overclocked a lot my GPUs, and unlocked some power limits by flashing custom GPUs BIOS (Vbios) years ago, and completly forget this until updating nvidia drivers to the first one that include CUDA 9.10 -> direct windows BSOD, or cannot install drivers at all, until I read somewhere that new drivers should be modified to accept power limit modification. Now I have no time anymore to overclock so I revert back all my GPUs to original Vbios.
Also on my 7GPUs rig, I have one 980Ti that refused to install drivers (this 980Ti is not vbios modified by me), so I've checked the model and manufacturer and found the original bios and flashed it : can now install drivers (387.92 and up) again.