https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #49 from Richard Biener <rguenth at gcc dot gnu.org> ---
I have a fix for the copy-headers issue. Afterwards:
tree copy headers : 0.03 ( 0%) 0.01 ( 3%) 0.03 ( 0%)
4763 kB ( 3%)
with the remaining worst offenders being
df live&initialized regs : 10.97 ( 53%) 0.02 ( 6%) 11.11 ( 52%)
0 kB ( 0%)
alias stmt walking : 1.85 ( 9%) 0.07 ( 21%) 2.08 ( 10%)
92 kB ( 0%)
and the DF time is all spent in RTL invariant motion:
loop invariant motion : 0.44 ( 2%) 0.00 ( 0%) 0.40 ( 2%)
151 kB ( 0%)
`- df live regs : 0.01 ( 0%) 0.00 ( 0%) 0.03 ( 0%)
0 kB ( 0%)
`- df reaching defs : 0.84 ( 4%) 0.02 ( 6%) 0.96 ( 5%)
0 kB ( 0%)
`- df live&initialized regs : 11.10 ( 54%) 0.01 ( 3%) 11.02 ( 52%)
0 kB ( 0%)
`- df use-def / def-use chains : 0.00 ( 0%) 0.00 ( 0%) 0.01 ( 0%)
0 kB ( 0%)
alias stmt waling is spent in FRE/DOM (that's expected):
tree FRE : 0.75 ( 4%) 0.01 ( 3%) 0.91 ( 4%)
2505 kB ( 2%)
`- alias stmt walking : 1.75 ( 8%) 0.02 ( 6%) 1.63 ( 8%)
92 kB ( 0%)
dominator optimization : 0.19 ( 1%) 0.01 ( 3%) 0.22 ( 1%)
5850 kB ( 4%)
`- tree operand scan : 0.00 ( 0%) 0.00 ( 0%) 0.02 ( 0%)
0 kB ( 0%)
`- alias stmt walking : 0.11 ( 1%) 0.01 ( 3%) 0.08 ( 0%)
0 kB ( 0%)
`- loop init : 0.00 ( 0%) 0.00 ( 0%) 0.01 ( 0%)
788 kB ( 0%)
`- tree CFG cleanup : 0.01 ( 0%) 0.00 ( 0%) 0.02 ( 0%)
0 kB ( 0%)
