https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125174
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed| |2026-05-06
Ever confirmed|0 |1
Status|UNCONFIRMED |NEW
--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Reproduces also on a Zen4 machine with -Ofast -march=x86-64-v3 -mtune=znver3
and --param ix86-vect-compare-costs 0 vs 1.
Most notable perf differences are that we now see
2.27% 12762 tonto_peak.amd6 libmvec.so.1 [.]
_ZGVbN2v_sin_sse4
1.36% 7618 tonto_peak.amd6 libmvec.so.1 [.]
_ZGVbN2v_cos_sse4
0.79% 4483 tonto_peak.amd6 libmvec.so.1 [.]
_ZGVdN4v_sin_avx2
vs.
0.72% 4072 tonto_base.amd6 libmvec.so.1 [.]
_ZGVdN4v_sin_avx2
the SSE4 ones do not appear. And we have
1.45% 8157 tonto_peak.amd6 libm.so.6 [.]
__sincos_fma ▒
1.36% 7700 tonto_base.amd6 libm.so.6 [.]
__sincos_fma
