Devices, network and instances of C4D

[atome451]

Hi Aoktar,

For some months, i often work on packshots with a lot of versions. I have several GPU on my main computer and on another one computer on the network. Then i'm doing the final renders in a render queue on one Cinema 4D instance and preparations of next renders in the LiveViewer (with one GPU activated) on another instance of Cinema 4D (started with -parallel parameter).

Globally, this works really fine but the management of devices, especially network slave, is a little bit complicated. It's not clear which parameter is global, which parameter is saved with the file or which parameter is specific to the C4D instance (if exists). Sometimes the network render is activated on one C4D instance, sometimes the Live Viewer "steal" the network on the other C4D instance. Often, when you open a new file, the network render is activated again on LiveViewer. You have to remember to verify all the time (if by sample you have to restart C4D) and, sometimes, the network is stolen by the Live Viewer by mistake. If it's the case, it seems it's not possible to verify or activate the network slave again on the render queue during the process (if i understand well or i didn't found where). Then you have to stop renders, etc.

I understand the utility of the Option > Network render in LV and the network option in the PV render preferences. They should stay like that. But maybe you should modify the priority (main priority on PV render by sample) and/or add an alert box when an action is going to steal a network render in a PV rendering process (with a yes or no). This way, you "steal" a network only if you want to.

But i'd also like to suggest as alternative to add a mangement space in the Octane settings > Device tab which have the max priority on all other parameters and which would lock the access to devices and network. This way, you define a setup in the morning, you lock this setup and it will not change, even if you have to restart an instance of C4D, until you unlocks it.

Is it possible to do that?

[bepeg4d]

Hi atome451,
maybe you can try the new option of 3.07 and create different daemons configurations, and load them when needed:
viewtopic.php?f=33&t=62395
ciao beppe

[atome451]

Thanks for the info, bepeg4d.

This seems indeed very useful for people who have a GPU park to manage. My case is rather simple so this is not a useful function for me.

My concern is more the management of local and networked GPUs for background final rendering on an instance of C4D while preparing other files with only one active GPU on another instance of C4D. So it's more the rationalization of this process within C4D that concerns me than management at Octane level. EDIT: With a mixed environment of local and networked GPUs. For only local GPU, the actual system is perfect.