I just
wanted to know why you want to have different
configurations.
So it is job based. Not a problem with any
client or artist.
As I can see now the main reason would be to render still images
with all GPUs.
And by
default you want to render with one GPU.
I have
a workaround if your GPU machines have a similar ammount of
RAM.
As you
already have two render configs, use the default "redshift" one for one GPU per
Thread.
Then
name the other renderer in the render config file for
example "redshift_oneGPU".
At
last, you have to make sure than the other threads do not get a
job.
The
workaround is to tell RR that the job takes that many memory that RR can only
start one of them.
And
not even a "small" one with the GPU per Thread.
If
your machines have for example 32 GB RM, then I assume they have about 29 GB
free memory.
Add this to the redshift_oneGPU render config:
RequiredMemory=0~28
and
this into the default one:
RequiredMemory=0~5
Result:
2x28
GB is larger than 32GB, 28+5 as well. They can never render
together.
3x5GB
fits into the free memory of the machine, 1x28 should as
well.
regards,
Holger Schönberger