public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug tree-optimization/113411] New: ABS<a>*ABS<b> can be simplified to ABS<a*b>
@ 2024-01-16 0:23 pinskia at gcc dot gnu.org
2024-01-16 8:13 ` [Bug tree-optimization/113411] " rguenth at gcc dot gnu.org
2024-01-21 9:58 ` pinskia at gcc dot gnu.org
0 siblings, 2 replies; 3+ messages in thread
From: pinskia at gcc dot gnu.org @ 2024-01-16 0:23 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113411
Bug ID: 113411
Summary: ABS<a>*ABS<b> can be simplified to ABS<a*b>
Product: gcc
Version: 14.0
Status: UNCONFIRMED
Keywords: missed-optimization
Severity: enhancement
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: pinskia at gcc dot gnu.org
Target Milestone: ---
Take:
```
static inline int abs(int a)
{
signed b = a;
if (a < 0)
b = -b;
return b;
}
int foo1(int a, int b)
{
return abs(a) * abs(b);
}
```
foo1 can be simplified down to `abs(a*b)`. Note I think this is only valid for
signed overflowable types and not valid with -fwrapv (but I could be wrong).
Maybe ok for real types with -ffast-math.
^ permalink raw reply [flat|nested] 3+ messages in thread
* [Bug tree-optimization/113411] ABS<a>*ABS<b> can be simplified to ABS<a*b>
2024-01-16 0:23 [Bug tree-optimization/113411] New: ABS<a>*ABS<b> can be simplified to ABS<a*b> pinskia at gcc dot gnu.org
@ 2024-01-16 8:13 ` rguenth at gcc dot gnu.org
2024-01-21 9:58 ` pinskia at gcc dot gnu.org
1 sibling, 0 replies; 3+ messages in thread
From: rguenth at gcc dot gnu.org @ 2024-01-16 8:13 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113411
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Ever confirmed|0 |1
Last reconfirmed| |2024-01-16
Status|UNCONFIRMED |NEW
--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
I think it's OK for reals with -fno-rounding-math.
It's valid for integers with or without undefined overflow, the corner
case to consider is
abs(-INT_MAX - 1) * abs (b)
where with undefined overflow the abs() is already invoking undefined
behavior. The problematical multiplication is (-INT_MAX-1) * -1 but
we cannot arrive at this from a defined input. For
abs(a*b)
the problematical input is -INT_MAX - 1 but we can't arrive at that
either (b == 1 with a == -INT_MAX-1 is undefined, b == -1 the same).
^ permalink raw reply [flat|nested] 3+ messages in thread
* [Bug tree-optimization/113411] ABS<a>*ABS<b> can be simplified to ABS<a*b>
2024-01-16 0:23 [Bug tree-optimization/113411] New: ABS<a>*ABS<b> can be simplified to ABS<a*b> pinskia at gcc dot gnu.org
2024-01-16 8:13 ` [Bug tree-optimization/113411] " rguenth at gcc dot gnu.org
@ 2024-01-21 9:58 ` pinskia at gcc dot gnu.org
1 sibling, 0 replies; 3+ messages in thread
From: pinskia at gcc dot gnu.org @ 2024-01-21 9:58 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113411
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Blocks| |56355
--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
PR 56355 is the same but for floating point already ...
Referenced Bugs:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56355
[Bug 56355] abs and multiplication
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2024-01-21 9:58 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-01-16 0:23 [Bug tree-optimization/113411] New: ABS<a>*ABS<b> can be simplified to ABS<a*b> pinskia at gcc dot gnu.org
2024-01-16 8:13 ` [Bug tree-optimization/113411] " rguenth at gcc dot gnu.org
2024-01-21 9:58 ` pinskia at gcc dot gnu.org
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).