Hi Everyone,
The makefile uses -O2 as the default optimization level.
The source code can support -O3 as the optimization level. In fact, we support -O5, -Os and -Ofast. We regularly test them in cryptest.sh (
https://github.com/weidai11/cryptopp/blob/master/cryptest.sh).
At -O3, GCC and other compilers will vectorize, meaning they will use SIMD units, like SSE and NEON. When the compiler takes an opportunity it is often a non-trivial speedup.
My question is, should we switch to -O3 as the default? Or maybe, is anyone opposed to doing so?
Jeff