I know wrong forum but I think this is some content some of you might find useful.
Took some time to benchmark resolve, new AI tools and heavy plugins that colorists love to use have slowed down our mac studio and so i tried it with our most beefy machine and some tests, just writing it all down , obviously grain of salt. but this project has it all, depthmaps, magic masks, neat video, all the DCTLs, Filmbox, you name it. timeline length is 3min8s , most 4608x3164 res , some shots extended for 9:16 to like 7K height, just like a “normal” project for us.
Resolve version 20.2.3 , Neatvideo V6, filmbox Pro.
Machine1 aka “Blackwell” → Window11 , 256Gb ram , Threadripper 9970X, RTX6000 blackwell , pcieG5 NVME
Machine2 aka “MacStudio” → Seqoia , 196GB ram, M2 Ultra (full-full config).
Network: 10Gbit
NAS: 24x NVME with 25Gbit Backbone connection, each machine was rendered after each other not at the same time
Test01:
i just tested pure Filmbox pro performance on both, with a generated source rendering to local fast disk (8bit dpx)
Blackwell: 41s
MacStudio: 1min32
Test2:
A realworld timeline reading DWAB exrs from network and writing DWAB EXRs back to network
Blackwell: 12min04s
MacStudio: 19min14s
Test03:
same timeline as above but this time reading uncompressed EXR (16bf) sources from local disk and rendering to 8bit DPX back to local disk.
Blackwell: 7min53s
MacStudio: 16min47s