We are doing some testing with a few Mac Mini M4 Pro workstations to be used as extra power on our farm when they are idle. Testing at the moment is with Cinema4D rendering with Redshift.
All path mapping is being handled by Deadline correctly, so when we set off a job we can see the Mac pick it up and launch C4D fine. The problem we are seeing at the moment, is that somewhere between Deadline, and the workstation, there is confusion about the M4 chip, and we don’t get proper allocation of the resources.
If I pull the full render command that Deadline is using, and run it in terminal directly on the Mac, then it launches C4D commandline and runs the render absolutely fine.
When I pass the job through from Deadline, it fails on allocating memory for the job. I’m seeing a difference in the worker logs, on how it sees the M4 Chip.
Local terminal run render
Deadline run render
It looks like the Deadline run process then gets in a state where as it doesn’t understand the Apple Silicon architecture, it allocated all of the memory to the CPU process, then when it gets to the GPU, there is no memory left to allocate. Any further in the worker logs have the device listed as “AMD Compatibility Mode” and it gets stuck in a loop of trying to run the task but then failing due to lack of memory.
The best I can manage with the GPU affinity is to disable the GPU portion, which gets it running albeit on CPU only and crawling along. Deadline says that the CPU affinity is only Window and Linux.
Cinema4d - 2025.2.1
Redshift - 2025.3.0
Deadline - 10.4.0
thanks,
Ed