https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115825
--- Comment #9 from Richard Biener <rguenth at gcc dot gnu.org> ---
It's odd that you see
Loop 1 iterates 7 times.
Loop 1 iterates at most 7 times.
Loop 1 likely iterates at most 7 times.
Estimating sizes for loop 1
BB: 3, after_exit: 0
size: 1 v ={v} i_10;
size: 0 i.0_1 = (unsigned char) i_10;
Induction variable computation will be folded away.
size: 1 _2 = i.0_1 + 254;
Induction variable computation will be folded away.
size: 0 i_7 = (char) _2;
Induction variable computation will be folded away.
size: 1 ivtmp_3 = ivtmp_4 - 1;
Induction variable computation will be folded away.
size: 2 if (ivtmp_3 != 0)
Exit condition will be eliminated in peeled copies.
Exit condition will be eliminated in last copy.
Constant conditional.
BB: 5, after_exit: 1
size: 5-4, last_iteration: 5-4
Loop size: 5
Estimated size after unrolling: 5
while I correctly see an estimated size after unrolling of 8 ... note this
still then runs into the 2/3 reduction.
Ah, the dumping is because trunk no longer does that 2/3 scaling in
estimated_unrolled_size.
So I'm still of the opinion that we should restrict the 2/3 scaling to
a specific set of stmts (or do the reverse and exclude another set,
like in this case volatile accesses).
