“However, I discovered a simple update to a widely used technique that makes it possible to run twice as fast as existing solutions on the Apple M1.”
CRC32 is sixty years old and sits inside zip, PNG, ext4 and Ethernet. The dedicated ARM instruction tops out around 25GB/s in theory. Adapting Intel’s 2009 carry-less multiply method to the M1’s wide core doubles the best existing code. Decades of optimisation still left half the speed on the table.