Which type of CPU cores are most active during Visual Look Up? How their frequencies and active residencies change? How demanding is it?
powermetrics
If Visual Look Up is so easy and low-power for Apple silicon Macs, maybe Tahoe’s new Foundation Models will prove more challenging, and wake up the neural engine.
Using powermetrics and log entries, a single image was processed on an M4 Pro, with content analysis and object recognition and look-up. How much power and energy did that use?
Tuning your Mac for performance can be a good investment of time. Beware of general benchmarks, though, and develop your own objective measurements. Then identify the rate-limiting step methodically, so you can address that.
How macOS controls CPU P core cluster frequency according to the cluster total active residency, in synthetic in-core tests, compression and when running virtual machines.
A matrix multiplication test appears to be run on the AMX matrix co-processor, and behaves differently from in-core tests. And what Power modes really do.
How could you study how Apple silicon CPU cores are used to run code? Comparisons between Activity Monitor, Xcode Instruments, and powermetrics.
How to compare an undocumented if not secret co-processor? Using different tests that use very high power, and can result in strange patterns of core allocation. So how does the M3 Pro fare here?
Comparison with M1 variants, energy use with comparison between M3 Pro and Max, virtualisation, Game Mode, vector processing and matrix co-processing – all in summary.
Assessing throughput using tests of fast Fourier transforms and sparse Cholesky factorisation from the Accelerate library. Is there an AMX there?
The Eclectic Light Company 