public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug tree-optimization/103633] New: Missed popcount recognition
@ 2021-12-09 15:29 acoplan at gcc dot gnu.org
  2021-12-09 15:42 ` [Bug tree-optimization/103633] " pinskia at gcc dot gnu.org
  2022-10-10 18:10 ` pinskia at gcc dot gnu.org
  0 siblings, 2 replies; 3+ messages in thread
From: acoplan at gcc dot gnu.org @ 2021-12-09 15:29 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103633

            Bug ID: 103633
           Summary: Missed popcount recognition
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: acoplan at gcc dot gnu.org
  Target Milestone: ---

For:

int f1 (unsigned long b) {
    int c = 0;
    while (b) {
        b &= b - 1;
        c++;
    }
    return c;
}

int f2(unsigned long x) {
    int c = 0;
    for (; x; x >>= 1) {
        if (x & 1)
            c++;
    }
    return c;
}

GCC recognizes f1 as popcount but not the more naive f2.

^ permalink raw reply	[flat|nested] 3+ messages in thread

* [Bug tree-optimization/103633] Missed popcount recognition
  2021-12-09 15:29 [Bug tree-optimization/103633] New: Missed popcount recognition acoplan at gcc dot gnu.org
@ 2021-12-09 15:42 ` pinskia at gcc dot gnu.org
  2022-10-10 18:10 ` pinskia at gcc dot gnu.org
  1 sibling, 0 replies; 3+ messages in thread
From: pinskia at gcc dot gnu.org @ 2021-12-09 15:42 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103633

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu.org
   Last reconfirmed|                            |2021-12-09
           Severity|normal                      |enhancement
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
We don't recongize this either:
int f21(unsigned long x) {
    int c = 0;
    for (; x; x >>= 1) {
        c+= (x & 1);
    }
    return c;
}

Confirmed.

^ permalink raw reply	[flat|nested] 3+ messages in thread

* [Bug tree-optimization/103633] Missed popcount recognition
  2021-12-09 15:29 [Bug tree-optimization/103633] New: Missed popcount recognition acoplan at gcc dot gnu.org
  2021-12-09 15:42 ` [Bug tree-optimization/103633] " pinskia at gcc dot gnu.org
@ 2022-10-10 18:10 ` pinskia at gcc dot gnu.org
  1 sibling, 0 replies; 3+ messages in thread
From: pinskia at gcc dot gnu.org @ 2022-10-10 18:10 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103633

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
A few extra testcase:
```
unsigned short fs(unsigned int execs)
{
  unsigned i;
  unsigned short num_algorithms = 0;
    for (i=0; i<32; i++) {
        if ((1<<i) & execs)
          num_algorithms++;
    }
  return num_algorithms;
}

unsigned short fi(unsigned int execs)
{
  unsigned i;
  unsigned int num_algorithms = 0;
    for (i=0; i<32; i++) {
        if ((1<<i) & execs)
          num_algorithms++;
    }
  return num_algorithms;
}

```
These testcases comes from coremarks but from the startup code so it has no
performance impact ...
You will also notice that fi has better code gen than fs with the vectorizer.

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2022-10-10 18:10 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-12-09 15:29 [Bug tree-optimization/103633] New: Missed popcount recognition acoplan at gcc dot gnu.org
2021-12-09 15:42 ` [Bug tree-optimization/103633] " pinskia at gcc dot gnu.org
2022-10-10 18:10 ` 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).