[Bug tree-optimization/36228] New: redundant runtime check while vectorizing

[Bug tree-optimization/36228] New: redundant runtime check while vectorizing

[razya at il dot ibm dot com]

/* { dg-do compile } */
/* { dg-options "-O3 -c" } */

#define COLS         8
#define ROWS         8

int
t_run_test(void);

int
t_run_test()
{
     int k_1,i_1, j_1;
     static signed char f_1[ROWS][COLS] ;
     static long F_1[ROWS][COLS] ;
     long cosMatrixA[ROWS][COLS] ;

     for( k_1 = 0 ; k_1 < COLS ; k_1++ )
        {
            for( i_1 = 0 ; i_1 < ROWS ; i_1++ )
            {
                for( j_1 = 0 ; j_1 < COLS ; j_1++ )
                    F_1[i_1][j_1] += f_1[i_1][k_1] * cosMatrixA[k_1][j_1] ;
            }
        }

  return 0;
}

A redundant check for aliasing is created.
Looking at the vectorizer's dump, we see:
versioning for alias required: can't determine dependence between
F_1[i_1_41][j_1_42] and F_1[i_1_41][j_1_42]

Tested on Power6.


-- 
           Summary: redundant runtime check while vectorizing
           Product: gcc
           Version: 4.4.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: razya at il dot ibm dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36228

[Bug tree-optimization/36228] redundant runtime check while vectorizing

[rguenth at gcc dot gnu dot org]

------- Comment #1 from rguenth at gcc dot gnu dot org  2008-05-13 13:40 -------
I belive this is a known issue with the data-dependence analyzer?  (I never
understood why it doesn't recognize identical references...)


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |spop at gcc dot gnu dot org
           Severity|normal                      |enhancement
           Keywords|                            |missed-optimization


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36228

[Bug tree-optimization/36228] redundant runtime check while vectorizing

[spop at gcc dot gnu dot org]

------- Comment #2 from spop at gcc dot gnu dot org  2008-05-14 03:59 -------
Subject: Re:  redundant runtime check while vectorizing

How costly would it be if we do as a first data dependence test a call
to operand_equal_p on the array references?  That certainly would
avoid lot of computations downstream in the data dependence test, but
would be applied as an initial test to all dependence relations.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36228

[Bug tree-optimization/36228] redundant runtime check while vectorizing

[sebpop at gmail dot com]

------- Comment #3 from sebpop at gmail dot com  2008-05-15 21:55 -------
Subject: Re:  redundant runtime check while vectorizing

Here is a patch for this:
> a first data dependence test a call to operand_equal_p on the array references
Tested with vect.exp and tree-ssa.exp.
I will send another email to gcc-patches with the patch once it passes
the tests on gccfarm.

Sebastian


------- Comment #4 from sebpop at gmail dot com  2008-05-15 21:55 -------
Created an attachment (id=15643)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=15643&action=view)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36228

[Bug tree-optimization/36228] redundant runtime check while vectorizing

[rguenth at gcc dot gnu dot org]

------- Comment #5 from rguenth at gcc dot gnu dot org  2008-05-15 22:01 -------
operand_equal_p is reasonably cheap.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36228

[Bug tree-optimization/36228] redundant runtime check while vectorizing

[spop at gcc dot gnu dot org]

------- Comment #7 from spop at gcc dot gnu dot org  2008-05-16 16:46 -------
Fixed.


-- 

spop at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36228