[Bug other/17103] New: Opportunity to replace multiply with shift/add sequence

[Bug other/17103] New: Opportunity to replace multiply with shift/add sequence

[gcc-bugzilla at gcc dot gnu dot org]

Description:
A non-optimal code sequence is illustraded.    Duplicate using gcc 3.5 and
command line:

gcc -O3 -m32 -c test.c


Testcase:
long foo(long x)
{
 return (x * 258);
}

Assembly:
Currently, gcc 3.5 generates the following:

foo:
      mulli 3,3,258
      blr

It would be more efficient to use a shift/add sequence:

foo:
      or 0,3,3
      rlwinm 3,3,8
      add 0,0,0
      add 3,3,0
      blr




-- 
           Summary: Opportunity to replace multiply with shift/add sequence
           Product: gcc
           Version: 3.5.0
            Status: UNCONFIRMED
          Severity: enhancement
          Priority: P1
         Component: other
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: steinmtz at us dot ibm dot com
                CC: gcc-bugs at gcc dot gnu dot org,steinmtz at us dot ibm
                    dot com
 GCC build triplet: powerpc64-linux
  GCC host triplet: powerpc64-linux
GCC target triplet: powerpc64-linux


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17103

[Bug middle-end/17103] Opportunity to replace multiply with shift/add sequence

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-08-19 15:34 -------
Confirmed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
          Component|other                       |middle-end
     Ever Confirmed|                            |1
           Keywords|                            |missed-optimization
   Last reconfirmed|0000-00-00 00:00:00         |2004-08-19 15:34:29
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17103

[Bug target/17103] Opportunity to replace multiply with shift/add sequence

[falk at debian dot org]

------- Additional Comments From falk at debian dot org  2004-08-19 15:46 -------
This works on Alpha, so I guess it is something target specific, like
imprecise insn costs.


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|middle-end                  |target


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17103

[Bug target/17103] Opportunity to replace multiply with shift/add sequence

[dje at gcc dot gnu dot org]

------- Additional Comments From dje at gcc dot gnu dot org  2004-09-12 20:37 -------
On POWER4, the default processor, the latency of mulli relative to shifts and
adds makes mulli a better choice than synthesizing the multiply by constant. 
When tuning for other PowerPC processors, such as POWER3, the synthesized
sequence is generated.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17103