Ed, we do not have a definitive answer for this and really have no easy way to test, but if we think of some way to test, will attempt it.
I am using several headless machines (cin -d 1200) as a render farm.
My understanding is that generally headless machines would not have any graphical libraries installed or X for that matter. At any rate, the vdpau and/or vaapi software would need to be installed for usage on each client.
Question :
If the master node is using an accelaration (HW) do the slave nodes automatically use the said acceleration ?
Without testing, we can not say for sure, but the $HOME/.bcast5/Cinelerra_rc file is used for all of the nodes, master and clients.
If that file has "USE_HW_DEV vaapi" or "USE_HW_DEV vdpau", conceivably the clients would also use the designated hardware device if the graphics board can handle it. if no but is a matter of local configuration for a slave where can I
set it up ?
I do not know if there is any way to set this up or to vary between vaapi or vdpau on different clients. The one thing you
can try is to set an environment variable for specific clients like the following and hope it uses that specific Cinelerra_rc CIN_CONFIG="/pathname>/<filename_such_as_.bcast5>" {path}cin -d port# where that specific Cinelerra_rc has "USE_HW_DEV=vdpau" or "USE_HW_DEV=vaapi" appropriately. Alternatively, an environment variable can be set and tried. For example: export CIN_CONFIG=/tmp/.bcast5 or even export IN_HW_DEV=vdpau or export CIN_HW_DEV=vaapi It looks like this will take some effort to figure out and if you do find something, please let us know. Phyllis/GG