OctaneRender™ 2018.1 for Houdini™ beta build 2018.1.0.3 RC3

[juanjgon]

Hey Juanjo,
thanks
I put NVLINK to test and it seams that octane crashes when I hit the VRAM capacity of one card - is VRAM pooling one of the features?
I simply created high-poly geo ... do you have any better idea to test if it is working, pls?
54093&hilit=nvlink#p354093

Hi,

To be honest, I'm not even sure if this feature is exposed in the plugins yet. Do you know what happens with Standalone if you try to render a huge scene? You could try to export it from Houdini using the scene exports functions.

Thanks,
-Juanjo

[juanjgon]

Yes its the octane OpenSubdiv feature.Any idea when this update is going to be released?
Thanks for the prompt reply.

The team is working hard to have it ready as soon as possible. It should be a matter of a few days.

Thanks,
-Juanjo

[Hampayon]

. Fixed an issue on Linux causing the computer to be identified differently if Houdini+Hython is being used. As a result a second license would have been required if Octane was
already run as Standalone or as different plugin.

Big thanks for this!

[zbynekkrulich]

Hi,
To be honest, I'm not even sure if this feature is exposed in the plugins yet. Do you know what happens with Standalone if you try to render a huge scene? You could try to export it from Houdini using the scene exports functions.

hey, ok, no problem - goldorak wrote it here viewtopic.php?f=33&t=69461
Unfortunately I can't test it in standalone since export from Houdini is locked due Indie version restrictions and standalone Octane can't see my license. I have Annual All-access subscription. Ehm

Best
zbynek

[ptunstall]

For both RC1 and RC3 I get this error in my linux shell when I try to connect to slaves:


CUDA error 700 on device 0: an illegal memory access was encountered
-> failed to deallocate device memory

I've tried nvidia and libcuda1 drivers 410.78 and 415.27 in linux. Do I need to use something newer than this?

I do not have this issues with your latest release of 4.

[juanjgon]

For both RC1 and RC3 I get this error in my linux shell when I try to connect to slaves:


CUDA error 700 on device 0: an illegal memory access was encountered
-> failed to deallocate device memory

I've tried nvidia and libcuda1 drivers 410.78 and 415.27 in linux. Do I need to use something newer than this?

I do not have this issues with your latest release of 4.

Hi,

So do you get this error on the slaves, or in the main Houdini session when you try to use the network rendering feature?

Thanks,
-Juanjo

[ptunstall]

This is the error I receive in the linux shell on the slave machines. Locally houdini/octane works fine in both RC1and RC3.

[WhyGee]

Hi,
I seem to be having trouble with cryptomattes generating random values for each frame. I've tried using all cryptomatte passes types but none seem to be consistent throughout the animation. Am I do something wrong?

Thanks
Yaniv

[juanjgon]

Hi,
I seem to be having trouble with cryptomattes generating random values for each frame. I've tried using all cryptomatte passes types but none seem to be consistent throughout the animation. Am I do something wrong?

Hi,

This problem can be related to the multi-thread scene extraction feature. If it is enabled the objects extraction order can change among the frames, and with it all the object and material IDs, so I'm afraid that you need to disable this option to render cryptomatte passes in animated sequences.

Thanks,
-Juanjo

[WhyGee]

great thanks alot!