------- Comment #9 from nbenoit at tuxfamily dot org 2009-12-16 11:06 -------
Here is a unified diff which focuses on the inner-loop exit conditions.
--- 442/convol.s
+++ r155286/convol.s
.L3:
movl (%edx), %ebx
- imull (%esi,%eax,4), %ebx
+ imull H(,%eax,4), %ebx
addl %ebx, %ecx
addl $1, %eax
subl $4, %edx
cmpl $511, %eax
- jg .L10
+ setle %bl
cmpl %edi, %eax
- jle .L3
-.L10:
+ setle -21(%ebp)
+ testb %bl, -21(%ebp)
+ jne .L3
movl -16(%ebp), %eax
movl %ecx, vH(,%eax,4)
-.L6:
+.L5:
L3 corresponds to the inner loop body.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42027
