Hi is there any way to control the order of the GPUs and/or which GPU is used to build the scene?
This is based on the following Problem:
I have 3 GPU Cards in my machine
2 x 970s ( lots of CUDA cores 4/3,5GB)
1 x 730 ( few CUDA cores 4GB)
(Monitors are connected to a non NVIDIA onboard GPU)
they are detected (consistently) as
[0] 970
[1] 730
[2] 970
DS seems always to use [0] as the device to build the scene what seem to use up some of the RAM
When I render Scenes that are allready near the RAM limit for the 970s i get Warnings in the log for device [0] and in the final statistic this device does a lot less iterations than the other 970. This significantly increase Render Time (nearly doubles it)
Is there a way to bring DS to use the 730 to build the scene. 1) it has more usable RAM 2) if i loose iterations on it it would have a much smaller impact on the render time.
(4.9 on a Win10 64 bit machine)