This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: mudflap problem
On Fri, Feb 04, 2005 at 01:05:52AM -0500, Doug Graham wrote:
> On Thu, Feb 03, 2005 at 04:40:12PM -0500, Doug Graham wrote:
> The test program parse.c is the same one I sent you a while back.
> Here are the numbers:
>
> neither: 0.43s user 0.13s system 101% cpu 0.553 total
> bounds: 7.01s user 0.28s system 99% cpu 7.297 total
> mudflap: 25.27s user 5.08s system 100% cpu 30.339 total
Forgot to mention: this is on a Fedora Core 3 system running on an Athlon
XP 3200+. I find it interesting that mudflap used almost as much system
CPU as bounds checking used total CPU. AFAIK, the only extra system CPU
that would be used by mudflap are the frequent calls to gettimeofday.
So I tried a run after setting -no-timestamps in MUDFLAP_OPTIONS and
got this:
mudflap: 19.51s user 0.56s system 99% cpu 20.075 total
Saved a bunch of user CPU too.
Getting closer.
--Doug.