https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102948
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords|needs-bisection |
CC| |marxin at gcc dot gnu.org
--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Hmm, I can only reproduce a 28s -> 32s regression between those two revs.
2fc2e3917f9c8:
runspec finished at Wed Oct 27 09:03:20 2021; 28 total seconds elapsed
274.28user 9.80system 0:26.92elapsed 1055%CPU (0avgtext+0avgdata
604912maxresident)k
0inputs+154168outputs (0major+4426618minor)pagefaults 0swaps
c16f21c7cf97ce:
runspec finished at Wed Oct 27 09:04:36 2021; 32 total seconds elapsed
308.73user 10.08system 0:30.37elapsed 1049%CPU (0avgtext+0avgdata
605356maxresident)k
0inputs+157344outputs (0major+4786067minor)pagefaults 0swaps
OK, so user time regressed more so the LNT numbers are wall-clock?
with -flto-partition=none I get
runspec finished at Wed Oct 27 09:13:45 2021; 215 total seconds elapsed
245.08user 5.85system 3:33.66elapsed 117%CPU (0avgtext+0avgdata
837776maxresident)k
0inputs+148856outputs (0major+1400075minor)pagefaults 0swaps
vs.
runspec finished at Wed Oct 27 09:19:56 2021; 243 total seconds elapsed
274.11user 6.07system 4:02.38elapsed 115%CPU (0avgtext+0avgdata
838792maxresident)k
0inputs+152056outputs (0major+1416810minor)pagefaults 0swaps
bisecting.
