https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109221
Hongtao.liu <crazylht at gmail dot com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |crazylht at gmail dot com --- Comment #5 from Hongtao.liu <crazylht at gmail dot com> --- (In reply to Andrew Pinski from comment #4) > With -ffast-math -mfpmath=387,sse (or -mavx512f instead of -mfpmath=387 as > there is a avx512f instruction too) added, ldexp gets inlined. Note, vscaless accept a float32 operand for exp which is int32 in ldexpf, there maybe some precision loss to convert an int32 to float32, that's why Ofast is needed here.