[sdiy] GCC vs Clang for Audio DSP

Eric Brombaugh ebrombaugh at gmail.com
Thu May 5 15:43:05 CEST 2022

On 5/5/22 04:43, Steve via Synth-diy wrote:
> Looking at the figures & seeing that Clang is faster at O0, I wonder
> whether it's perhaps doing some things that GCC is doing with -Og, which
> supposedly does "all optimizations that don't interfere with debugging".

That could well be - the exact nature of the fine-grained optimizations 
that clang does for the various -O settings isn't clear to me without 
more research. Notice for example that the clang results for -O2 and -O3 
are virtually identical, likely because the differences in individual 
optimizations that are lumped together for those don't affect my 
particular test case. Different code, or a different target would be 
needed to reveal what's changed between clang -O2 and -O3.

More information about the Synth-diy mailing list