GCC 12 enables autovectorization at -O2

Compiler geek inside me is excited about this change in GCC

https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=2b8453c401b699ed93c085d0413ab4b5030bcdb8

This is a ton of work to pull something like this and make it default. ARM chips with neon SIMD benefits and so do all modern chips. I will do some benchmarking with gcc and see how it compares at O2 with gcc11

1 Like