Serious Crash Error (URGENT help needed!!!)

[Draydin_r]

I'm having a serious issue with one of my renders. Every so often my screen will go black and a box will come up saying that my video card driver has recovered from an critical failure "Display Driver has stopped responding and has recovered", then one the video cards fail and the renders seemingly continues on regardless (unless both of the video cards fail). This only seems to occur when I render out extensive scenes (in this case exterior scenes with massive amounts of grass). I seemingly can render out interior scenes without much concern.

This occurs on both Lightwave Plugin and Octane Standalone. I am running 2 GeForce GTX TITANS with the Driver version 344.11.

If anyone can help me with this I would be extremely grateful.

Cheers,
Draydin_R

[rappet]

You might try install new driver 347.09 and see if that helps.
(do not forget to restart your PC after update)

[juanjgon]

Try this, You can manually change the settings to get a higher timeout via the registry editor:
- Launch the regedit (Start -> Run window -> regedit)
- Go to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers
- The key "TdrLevel" should be set to 3.
- The key "TdrDelay" is the delay in seconds. Maybe set it to 20 or 30.

Restart your system and try again.

-Juanjo

[abstrax]

Also run the Standalone installer at least once which will set the TdrDelay to 10s. After the installation you have to reboot Windows.

Background: If a kernel call takes longer than 0.5s or 2s (depending on the Windows version), Windows will kill the driver, which gives the error message you have reported.

[Draydin_r]

Hi guys,

Thanks for the response! So far I have reinstalled octane with the latest version then rebooted, and then adjusted Tdrlevel and TdrDelay settings and rebooted, and they have had no effect as the cards still crash, though slightly later than before.

Also, what did you mean by Background abstrax? Where you referring to the the two titan cards i have running background? Is there anything I can do to alleviate this issue or should I be having my cards looked at? Would updating my drivers help as suggested earlier by rappet (I wasn't ignoring your suggestion rappet, just making sure that i tried the suggestions of the Devs first ).

Eitherway is there anything else I can try to help me? I would be panicking more if this was affecting every render I have been doing, instead it has only been affecting scenes with massive numbers of instancing (mostly my exteriors). What I'm afraid of is this expanding into a bigger problem.

Here is the log for that session:
-----------------------------------------------------------------------------
Started logging on 09.02.15 10:53:57

OctaneRender version 2.21.1 (2210003)

CUDA error 999 on device 0: An unknown internal error occurred.
-> kernel execution failed (pt)
device 0: path tracing failed
CUDA error 999 on device 1: An unknown internal error occurred.
-> kernel execution failed (pt)
device 1: path tracing failed
-----------------------------------------------------------------------------

Please let me know if you guys have any suggestions.

Draydin_R

[juanjgon]

Perhaps you have a lot of instances overlapped over the same place, and the render becomes too slow to get the first sample within a reasonable time.

Can you please make two tests? First try to render the scene with half instances, and second with the same instances, but at half resolution.

Also I can see that you have have surfaces with displacement. Please, try to use lower displacement distance or disable the displacement to see if this could also be the problem.

-Juanjo

[smicha]

I'll try to focus on hardware side - I had similar issues with 580 which I RMA.

1. Is your 770 set as a primary display card? If so, is it inserted into the first (closest) to a CPU PCI express slot? Do you use more than one monitor? If so connect all of them to 770.

2. What PSU do you have? Is it less than 1000/1200W? Is it high end psu?

3. If titans are for rendering only there is no need to use 'priority' option. Use it for 770 (if used for rendering) and set it to low.

4. What is the voltage at which titans go? 1.16 or 1.2? Bumping up voltage to 1.2V may give better stability. You can monitor voltage in MSI Afterburner... What are temps and core clocks?

5. What is the brand of your titans? I had problems with gainward 580. With asus (and evga 680) - never. Do both of them fail? Test it with a single titan on in Octane.

6. Do you overclock your cpu with offset? Or do you have power saving settings off in bios? Did you try to set manually voltages for your cpu?

[face_off]

Have you overclocked the cards? If so, try setting them back to the normal clock speeds.

Paul

[smicha]

Have you overclocked the cards? If so, try setting them back to the normal clock speeds.

Paul

+1

This is very important especially when you reach high temps and have a poorly ventilated case.

[Draydin_r]

No over clocking on either cards and they never reach temps past 80 C.