@mikeywestcott, can you please provide more specifics? The Sony mirrorless cameras have many different formats of XAVC: H.264, HEVC, 8-bit, 10-bit 4:2:0, 4:2:2, XAVC-SI All-I, etc. They each have different performance characteristics during NLE playback. However, FCP 11.0 should not be that slow on a two-camera edit (excluding effects). There may be some narrow issues with certain effects unique to FCP 11.0 we have not yet narrowed down fully, but since effects can slow down any machine and any NLE, we'd need more information.
To rule out effects, duplicate your project, then load the duplicate project, select all timeline clips with CMD+A, and remove all effects with Edit>Remove Effects. Then try the multicam playback and export performance. If that makes a huge difference, it is some effect.
I just edited in FCP 11.0 a 27.5-minute two-camera interview shot by two Sony A7RIIIs using 4/29.97 8-bit 4:2:0 XAVC-S. I did this on both an M1 Ultra Mac Studio (128 GB, 64 GPU cores), M1 Mac MacBook Pro 16 (64GB, 32 GPU cores), and an M4 Pro Mac Mini 48GB, 20 GPU cores), both running Sequoia 15.1.1. I used basic grading with color wheels, and the timeline was not rendered to cache. The content was on the internal SSD on all three machines.
With the angle viewer up, the viewer in Better Quality mode, no background rendering and no proxies or optimized media, the 4k timeline scrubbing performance was good on all machines. With all the dropped frame indicators active, I got no dropped frames.
I evaluated the viewer update rate on all three machines during various fast forward and fast reverse modes using the JKL keys, also when dragging the playhead and using the skimmer. In general it was good, although not quite as slippery fast as using optimized media.
My impression is the M4 Pro Mac Mini was equally responsive on multicam viewer update rate as the M1 Ultra, which in turn was somewhat more responsive than the M1 Max. I attribute that to the M1 Ultra having two parallel decode accelerators vs the M1 Max, and the M4 Pro's single decode engine being improved in the M4 series.
I then tested the export time using the File>Share>Export File>Settings, Format: Computer, Video Codec: H.264 Single-pass (Faster), Resolution 3840 x 2160, and 'Allow Export Segmentation' enabled (M1 Ultra & Max). It exported the 27.5 minute interview in 4 min 33 sec (6x real time) vs M1 Max at 8 min 15 sec (3.3x real time), and the M4 Pro at 15 min 51 sec (1.73x real time). I attribute the faster performance of the M1 Ultra and Max machines to parallel segmented encoding. The M1 Ultra has four hardware encoders, the M1 Max has two and the M4 Pro has one.
