What kind of increased performance are people on Arm64 computers getting with the new Arm64 VCV Rack 2.4.0? I suspect a lot of people might get better performance on Mac M1 and M2 systems in the new VCV build.
I have the original base level MacBook Air M1. I don’t see much difference in “typical” patches. But in extreme cases the ARM version can run patches that x86 cannot.
For example, my VCP#75 AirWindows challenge patch with 81 instances of AirWindows cannot run on x86 version - it pegs the max and average CPU usage above 220%. But the ARM version handles it with ease at 38% / 42%. Also I can run configured at 60 fps with actual output at 57 fps.
Another example is an FM patch where I was experimenting running at high sample rates to see if I could minimize aliasing. On the PC I could go up to 384kHz. Put on x86 M1 I was limited to 96kHz. Now with the Arm version I can go to 384kHz on the mac.
One other thing I notice is the Arm version seems to better support multiple threads. With the x86 version multiple threads never helped, and often prevented the patch from working. But the ARM version is able to run all my tested patches with multiple threads. I haven’t yet run across a patch where multiple threads is beneficial, but I suspect I will eventually.
2 Likes
indeed, I just retested it, on my previous 12 core Ryzen it was stuttering no matter how many cores I gave it, and on this here M1 Mini in arm64 Rack it rides at 37.5% on a single thread. Actually adding a thread makes it use 44% CPU. So far I have not seen any benefit of raising thread count above 1.