I used QEMU through Podman machine in the end, but the performance is terrible. Even on MacOS side QEMU does not support Rosetta 2. Docker for Desktop supports and gives significant boost, but I don’t want to use it.
I love the battery life and the keyboard is much better than the old one. It feels snappy running native code but I just know it's going to cause me more issues than it's worth.
Is Rosetta 2 that much faster than QEMU's own translation? Why?
Oh yes. Based on the benchmarks I've seen it's a good 30-40% faster on compute workloads. I assume that the pre-transpilation of the binaries is advantageous relative to having to virtualize the CPU.
† actually not measured, but compared to the order of magnitude on the other side it's about that, eyeballed against the same code on aarch64.
Among other things, Rosetta 2 is able to use the CPU's special TSO memory ordering mode, which I think QEMU's TCG cannot use so has to use barrier instructions instead (MTTCG), or run all the virtualised CPU threads on a single host core (obviously slower for parallel workloads).