Octane Crash with 3 GPUs

[blixen8]

Hi all,

I wondered if I could reach out to you for some help/advice.

I’ve got a custom workstation :-

Fractal Design Define R6C Chassis
Corsair HX1200 – 1200W (Platinum efficiency) PSU
ASUS WS X299 SAGE Mainboard (latest BIOS – 2002 from 18/10/2019)
3x ASUS “TURBO” Blower style RTX 2080 Ti cards (various GR and studio drivers tested)
1x Samsung 1TB 970 EVO Plus M.2 NVMe SSD
1x Western Digital 4TB “Ultrastar” HDD
Intel Core i9-9900X (10x 3.50GHz - 19.25MB Cache) w/ HT CPU
64GB (4x 16GB) DDR-4 3000MHz Corsair Vengeance LPX (this has been replaced)
Windows 10 Professional
Microsoft Wireless Keyboard and Mouse


Essentially, I’ve experienced some relatively frequent FREEZES / blue screens. I’m not working any differently to how I would on my old workstation.

The computer will just hang then crash when the IPR is running and I'm working away in the texture node editor. Sometimes just when the IPR is running.

This happens in different scenes and in Lightwave 2018, LW 2019, Octane4 and Octane 2019.

I can’t reproduce the fault all of the time and it doesn’t happen when I’m doing anything else on the system either. I can browse the internet, work in Photoshop quite happily. I can get a complete freeze every couple of days.

To troubleshoot so far, we have updated my BIOS to the latest version and changed the RAM to be on the safe side.

Tried dissabeling Denoising in GPU prefs.

Using 441.28 studio drivers but tried various other drivers, 417, 440 e.t.c.

If I remove one of the GPU's and run on two the problem seems to go away. If I simply disable one of the GPU's in Octane prefs the problem seems to go away.

I've replaced the GPU's to confirm it's not the GPU's.

GPUz readings are normal for GPU temp and Power.

Is there something I should be doing to have all three cards running?

Does anyone have any other experiences of similar behaviour or practical troubleshooting steps I might try?


For what it is worth my bluescreens are always the same and don’t particularly reference anything helpful :-

It’s hal.dll+1848

ntoskrnl.exe ntoskrnl.exe+1eeac1 fffff802`35e00000 fffff802`368b6000 0x00ab6000 0x4efcf7a9 29/12/2011 23:28:41 Microsoft® Windows® Operating System NT Kernel & System 10.0.18362.535 (WinBuild.160101.0800) Microsoft Corporation C:\WINDOWS\system32\ntoskrnl.exe
hal.dll hal.dll+1848 fffff802`368b6000 fffff802`36959000 0x000a3000 0x1cd3fb4f Microsoft® Windows® Operating System Hardware Abstraction Layer DLL 10.0.18362.387 (WinBuild.160101.0800) Microsoft Corporation C:\WINDOWS\system32\hal.dll

Your help would be very much appreciated. Thanks in advance!

[juanjgon]

This problem could be related to the PLX chipset included in your mainboard. Check the fist post in this topic, for example:
viewtopic.php?f=9&t=70934

Thanks,
-Juanjo

[blixen8]

Thanks Juanjgon,

I'll look into that.

[blixen8]

Just tried the MSI option but no joy, still crashes when all three cards are enabled.

[juanjgon]

This is a long shot, but another problem could be the PSU, not able to power all the three GPUs perhaps because it has some kind of problem.

Thanks,
-Juanjo

[blixen8]

I’ve got a HX1200 Corsair PSU – it’s Platinum efficiency (so 1150Wish) and the cards have a rated TDP of 285W each (they’re ASUS TURBO versions). So even loosely with a 165W TDP CPU (No Overclock)

3* 285W = 855W + 165W = 1020W

I realise there will be fluctuations, but likewise, everything won’t be full pelt, all of the time.

There is nothing else in the system that really draws any real power.

[juanjgon]

Have you checked the latest 441.66 STUDIO drivers?

Thanks,
-Juanjo

[blixen8]

No difference with the 441.66 drivers.
Going to change the PSU to a more powerfull unit, see if that works.