One aspect we’ve never really had the opportunity to test is exactly how good Apple’s cores are in terms of memory bandwidth. Inside of the M1, the results are ground-breaking: A single Firestorm achieves memory reads up to around 58GB/s, with memory writes coming in at 33-36GB/s. Most importantly, memory copies land in at 60 to 62GB/s depending if you’re using scalar or vector instructions. The fact that a single Firestorm core can almost saturate the memory controllers is astounding and something we’ve never seen in a design before.
Compared to the Intel contemporary designs, the Apple M1 is able to showcase a performance leap ahead of the best the company has to offer, with again a considerable strength in the FP score.
Considering that most of the people with scientific computing / data sciencey workloads these days are working off Macbooks, some pretty exciting stuff.
Is anyone on here planning on getting one of these fancy new ARM Macs, at least to make sure stan builds and runs correctly (though I understand there are already some problems with Big Sur even on X86)? Would also be fun to see model compiliation times and single-chain benchmarks with all these claims about the M1’s lead in single-core performance.