Does anyone here have a good source for how fast vector instructions (AVX2 and the like) will compute under Rosetta2 on the Arm-based Macs vs the current x86 Macs?