[Bug tree-optimization/16962] loop unrolling with x86-64 asm not efficient

["falk at debian dot org" <gcc-bugzilla at gcc dot gnu dot org>]

------- Additional Comments From falk at debian dot org  2004-08-10 14:55 -------
Well, the fundamental problem is independent of assembly. Here's a much simpler
example:

void f(int *p, int l) {
    int i;
    for (i = 0; i < l; ++i)
        p[i] = 0;
}

With 3.4 and also with an oldish lno, on Alpha I get

$L36:
        stq $31,0($3)
        lda $5,1($5)
        lda $3,8($3)
$L35:
        stq $31,0($3)
        lda $5,1($5)
        lda $3,8($3)
...

while the offsets could easily be constant folded if there were no jumps into
the loop.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16962