https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80399
Bug ID: 80399
Summary: Premature optimization with unsigned
Product: gcc
Version: 7.0
Status: UNCONFIRMED
Keywords: missed-optimization
Severity: normal
Priority: P3
Component: middle-end
Assignee: unassigned at gcc dot gnu.org
Reporter: pinskia at gcc dot gnu.org
Target Milestone: ---
Take the following two functions:
unsigned f(unsigned t)
{
t = (t - 1)*16;
t = t/16;
return t;
}
unsigned g(unsigned t)
{
t = t - 1;
t = t*16;
t = t/16;
return t;
}
They should produce the same code but they don't.
On aarch64 for an example we produce:
f:
mov w1, 268435455
add w0, w0, w1
and w0, w0, w1
ret
....
g:
sub w0, w0, #1
and w0, w0, 268435455
ret
g is better than f here since we have only one dependent instruction instead of
2.
