Pascal TitanX + 3.04.3 Problem with Lightwave 2015.3

Thu Nov 17, 2016 4:56 am

Lewis wrote:I have 2 TitanX Maxwels and 2*GTX 1080 so in case you guys need help with testing send me that scene so i can try Maxwell, Pascal and Combo of both.

cheers

---------------------------------------------
If you can test it, please let me know your e - mail address.

I will send you a file.

Thu Nov 17, 2016 8:39 am

Finally, I've had this problem last night rendering the sequence from the frame 1170, in the frame 1174, so yes, the problem is there.

02:21:44 (8504.26) * OCTANE API MSG: CUDA error 700 on device 0: an illegal memory access was encountered
02:21:44 (8504.26) * OCTANE API MSG: CUDA error 700 on device 0: an illegal memory access was encountered -> kernel execution failed(report)
02:21:44 (8504.26) * OCTANE API MSG: CUDA error 700 on device 0: an illegal memory access was encountered
02:21:44 (8504.26) * OCTANE API MSG: CUDA error 700 on device 0: an illegal memory access was encountered -> failed to launch kernel(ptBrdf2)
02:21:44 (8504.26) * OCTANE API MSG: device 0: path tracing kernel failed
02:21:44 (8504.26) * OCTANE API MSG: device 0: detected an error on render device! trying to recover...

Now the big issue is going to be try reproduce it. I am going to export the scene to .orbx to try to get the crash in Standalone.

Thanks,
-Juanjo

Thu Nov 17, 2016 8:42 am

Aha so good news. That "illegal memory" thing is bug i was experiencing often in 3.x also.

Let's hope it does same in standalone.

Thanks for update Juan

Thu Nov 17, 2016 9:16 am

juanjgon wrote:Finally, I've had this problem last night rendering the sequence from the frame 1170, in the frame 1174, so yes, the problem is there.

02:21:44 (8504.26) * OCTANE API MSG: CUDA error 700 on device 0: an illegal memory access was encountered
02:21:44 (8504.26) * OCTANE API MSG: CUDA error 700 on device 0: an illegal memory access was encountered -> kernel execution failed(report)
02:21:44 (8504.26) * OCTANE API MSG: CUDA error 700 on device 0: an illegal memory access was encountered
02:21:44 (8504.26) * OCTANE API MSG: CUDA error 700 on device 0: an illegal memory access was encountered -> failed to launch kernel(ptBrdf2)
02:21:44 (8504.26) * OCTANE API MSG: device 0: path tracing kernel failed
02:21:44 (8504.26) * OCTANE API MSG: device 0: detected an error on render device! trying to recover...

Now the big issue is going to be try reproduce it. I am going to export the scene to .orbx to try to get the crash in Standalone.

Thanks,
-Juanjo

Good News.

I am 1179 Frame Error

Maxwell TitanX, which has no rendering problems, is also upgraded to 3.04.3, and rendering in the scene will cause the LW 2015.3 program to crash.

Mon Nov 21, 2016 2:26 am

juanjgon wrote:Finally, I've had this problem last night rendering the sequence from the frame 1170, in the frame 1174, so yes, the problem is there.

02:21:44 (8504.26) * OCTANE API MSG: CUDA error 700 on device 0: an illegal memory access was encountered
02:21:44 (8504.26) * OCTANE API MSG: CUDA error 700 on device 0: an illegal memory access was encountered -> kernel execution failed(report)
02:21:44 (8504.26) * OCTANE API MSG: CUDA error 700 on device 0: an illegal memory access was encountered
02:21:44 (8504.26) * OCTANE API MSG: CUDA error 700 on device 0: an illegal memory access was encountered -> failed to launch kernel(ptBrdf2)
02:21:44 (8504.26) * OCTANE API MSG: device 0: path tracing kernel failed
02:21:44 (8504.26) * OCTANE API MSG: device 0: detected an error on render device! trying to recover...

Now the big issue is going to be try reproduce it. I am going to export the scene to .orbx to try to get the crash in Standalone.

Thanks,
-Juanjo

Did you find the problem?

Did you find the problem?

In Maxwell TitanX + version 3.04.3, Maxwell is also a problem.

Mon Nov 21, 2016 7:06 am

I tried to load your scene but LW argues for so many missing files (textures)

. ?

Mon Nov 21, 2016 7:44 am

Lewis wrote:I tried to load your scene but LW argues for so many missing files (textures) . ?

-------------------------------------

I'll check again.

And I will send you another mail again.

Mon Nov 21, 2016 7:56 am

Lewis wrote:I tried to load your scene but LW argues for so many missing files (textures) . ?

A total of 187 image files were used.
The image I sent was 186 images.
One image "001 (22) .jpg" may be replaced by another one.
I do not know. I will mail the file again.

Mon Nov 21, 2016 7:59 am

True that one is fully missing (octane can't use instanced images), it also argues about files which are in folder like "Plaster_pointydirectional_512_d.tga" but loader can't see them without manual retargeting (50+ clicks to find it for every popup). I guess you used that on some model and it got saved as Absolute path instead relative so it's problem for location that's not locally/at your network.

Try to disable/hide your network drive/share temporarily and load this scene from different location and you will see how many times it asks about texture regardless of it being in folder of Package scene. Sadly LWO files still save lot of stuff as Absolute paths and Package scene don't go into LWO file to change that, it just collects data form multiple places.

BUT either way, important part is that Juan did replicate this problem so hopefully he can fix this BUG

.

Mon Nov 21, 2016 8:37 am

Lewis wrote:True that one is fully missing (octane can't use instanced images), it also argues about files which are in folder like "Plaster_pointydirectional_512_d.tga" but loader can't see them without manual retargeting (50+ clicks to find it for every popup). I guess you used that on some model and it got saved as Absolute path instead relative so it's problem for location that's not locally/at your network.

Try to disable/hide your network drive/share temporarily and load this scene from different location and you will see how many times it asks about texture regardless of it being in folder of Package scene. Sadly LWO files still save lot of stuff as Absolute paths and Package scene don't go into LWO file to change that, it just collects data form multiple places.

BUT either way, important part is that Juan did replicate this problem so hopefully he can fix this BUG .

----------------------------------------------------------------------------------
It's not that important, but I sent it again.