3 ms·
In a similar spirit, Apple seems to have made sure some critical OSX idioms were fast on the M1, perhaps even influencing their instruction set. Retaining and
by willvarfar 3y ago
In a similar spirit, Apple seems to have made sure some critical OSX idioms were fast on the M1, perhaps even influencing their instruction set.
Retaining and releasing an NSObject took ~6.5 nanoseconds on the M1 when it came out, comparing with ~30 nanoseconds on the equiv gen Intel.
In fact, the M1 _emulated_ an Intel retaining and releasing an NSObject fast than an Intel could!
One source: https://daringfireball.net/2020/11/the_m1_macs https://daringfireball.net/2020/11/the_m1_macs
- ithkuil 3y agoThe M1 emulation with Rosetta is actually dynamic recompilation so of you're measuring only that specific small section it's not surprising that Rosetta could have emitted optimal code for that instruction sequence
- lxgr 3y agoThat seems very similar to ARM introducing an instruction set that can be used to efficiently implement common Java idioms to me, just that x86 bytecode wasn't explicitly conceived as a VM target.