https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115025
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed| |2026-02-13
Status|UNCONFIRMED |NEW
Known to fail| |16.0
Ever confirmed|0 |1
--- Comment #13 from Richard Biener <rguenth at gcc dot gnu.org> ---
Manually doing the CH does not get to GCC 13 performance on trunk still on a
Zen2 system. The trunk loop below is slightly better than the one from 13,
I do wonder how the prologue to it matters so much.
60: 8d 7e 02 lea 0x2(%rsi),%edi
63: 89 c8 mov %ecx,%eax
65: 31 d2 xor %edx,%edx
67: f7 f7 div %edi
69: 85 d2 test %edx,%edx
6b: 74 c3 je 30 <is_prime+0x30>
6d: 83 c6 06 add $0x6,%esi
70: 44 39 c6 cmp %r8d,%esi
73: 73 13 jae 88 <is_prime+0x88>
75: 89 c8 mov %ecx,%eax
77: 31 d2 xor %edx,%edx
79: f7 f6 div %esi
7b: 85 d2 test %edx,%edx
7d: 75 e1 jne 60 <is_prime+0x60>
