Page 1 of 1

Network Render Issue (3.08-RC2)

Posted: Sat Mar 24, 2018 9:45 am
by Teddi
Hi.

I'm still having network render issues with my Live Viewer in 3.08-RC2 C4D.

I have two GPUs in my slave machine which are working fine but often my LV isn't using them (info shows 0/2 instead of 2/2).

It seems to work OK in Picture Viewer or rendering via the Render Queue.

I think the size of the scene is making a difference (heavier scene = more likely to drop slave cards).

Could this be a network issue?

This didn't happen with 3.07.

System info..

Win 10 (fully updated).
Latest Nvidia drivers (as of date of posting).
C4D R17 (up to date)
Main machine: i7-6850K, 64gb RAM, 2X 1080ti
Slave machine: i7-7700K, 32gb RAM, 2X 1080

Re: Network Render Issue (3.08-RC2)

Posted: Sat Mar 24, 2018 10:24 am
by aoktar
Network transfer operation may take longer time than render time. Did you tried with max maxsamples for testing?

Re: Network Render Issue (3.08-RC2)

Posted: Sat Mar 24, 2018 11:34 am
by Teddi
Thanks Aoktar.

Do you mean use max samples for a long render?

This issue happens with long renders in LV (e.g. 20 minutes because of high GI and fairly high samples of 4000).

The slave cards sometimes come online after a delay, but often they're not used at all.

Perhaps I could monitor my network traffic somehow and see if there's a bottleneck?

Re: Network Render Issue (3.08-RC2)

Posted: Sat Mar 24, 2018 11:52 am
by aoktar
There's not much control to interfere the network operations from plugins. Just activate netrender and select clients. Rest is up to core and standalone in slaves. You should check slave output, network and GPU/CPU usage. Also there was a log flag to debug the standalone in slaves. We can follow the flow by that.

Re: Network Render Issue (3.08-RC2)

Posted: Mon Mar 26, 2018 8:27 am
by Teddi
Thanks Aoktar.

Is there any particular method or software you'd recommend to monitor slave output, network and CPU/GPU usage that I can run while rendering in the LV?