Apple M1 ARM Performance With A 2020 Mac Mini
What better way of breaking in a new system by tossing NAMD on there to really stress things... With the existing NAMD macOS package running under Rosetta for dynamically translating x86_64 to ARM64, the performance acrually was an immediate surprise given how heavy this molecular dynamics software package tends to be on the processor. The Rosetta-based performance wasn't quite as fast as the Core i7 8700B Coffee Lake of the previous generation Mac Mini but well ahead of a Skylake MacBook Pro and obviously a old Haswell Mac Mini.
The performance of x86_64 binaries on macOS Big Sur with the Apple M1 really depend upon the code-base. With MAFFT builds were done of x86_64 under Rosetta and then native 64-bit ARM The native ARM binary on the M1 easily blew away the performance of the previous-generation Core i7 8700B Mac Mini while the Rosetta performance here was slower than even the Haswell era Mac Mini from several years ago.
When firing up many existing macOS packages that haven't yet adopted universal builds or targeted builds otherwise for Apple Silicon, for the most part the Rosetta performance though is quite admirable and at least comparable to the previous generation Core i7 Mac Mini.
Zstd compression under Rosetta was one of the biggest surprises with significantly better compression performance compared to any of the tested Intel Macs.