[Bug debug/91817] compile with -O3 is more-than-expectedly slower than -O2

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91817

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |RESOLVED
      Known to work|                            |5.3.1
         Resolution|---                         |FIXED

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
On x86_64-linux with GCC 8 I see

rguenther@murzim:/tmp> /usr/bin/time gcc-8 -S 1.i -m32 -g
0.39user 0.02system 0:00.42elapsed 99%CPU (0avgtext+0avgdata 90428maxresident)k
0inputs+3648outputs (0major+20612minor)pagefaults 0swaps
rguenther@murzim:/tmp> /usr/bin/time gcc-8 -S 1.i -m32 -g -O
1.77user 0.05system 0:01.82elapsed 100%CPU (0avgtext+0avgdata
141920maxresident)k
0inputs+5280outputs (0major+33933minor)pagefaults 0swaps
rguenther@murzim:/tmp> /usr/bin/time gcc-8 -S 1.i -m32 -g -O2
8.01user 0.08system 0:08.10elapsed 99%CPU (0avgtext+0avgdata
330268maxresident)k
0inputs+6272outputs (0major+86763minor)pagefaults 0swaps
rguenther@murzim:/tmp> /usr/bin/time gcc-8 -S 1.i -m32 -g -O3
8.19user 0.11system 0:08.31elapsed 99%CPU (0avgtext+0avgdata
332380maxresident)k
0inputs+6344outputs (0major+94164minor)pagefaults 0swaps

Same for GCC 7.

note GCC 4.6 is very old so it's somewhat pointless to complain about its
compile-time speed.  Nevertheless I can reproduce the slowness with that
version so I am assuming this issue has been fixed, in 5.3.1 at least.