------- Comment #2 from uros at kss-loka dot si 2005-10-24 06:53 -------
Some updated timings with gcc version 4.1.0 20051021 (experimental):
-O2 -march=pentium4 -fprefetch-loop-arrays:
user 0m17.805s
user 0m17.752s
user 0m17.744s
-O2 -march=pentium4:
user 0m17.750s
user 0m17.758s
user 0m17.838s
The main loop with -fprefetch-loop-arrays now looks:
.L2:
prefetcht0 buf+384(,%eax,4)
movl %eax, buf(,%eax,4)
addl $1, %eax
cmpl $10000000, %eax
jne .L2
And without prefetch:
.L2:
movl %eax, buf(,%eax,4)
addl $1, %eax
cmpl $10000000, %eax
jne .L2
--
uros at kss-loka dot si changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |RESOLVED
Resolution| |FIXED
Target Milestone|--- |4.1.0
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20748
