Patch: Tune "dir" a bit

[Volker Lendecke]

On Mon, Mar 25, 2013 at 04:14:32PM -0400, Scott Lovenberg wrote:
> I'm starting to think that's the case.  Or a wildly different version
> of gcc.  I'm going to compile a 3.5 tarball and see if the numbers are
> more in line.  It won't have RH's changes, but it should be close
> enough for comparison.

Just for the fun of it: Slightly different directory
listing, but otherwise same test:

Without mutexes:

real    0m25.421s
user    0m11.529s
sys     0m5.912s

With mutexes:

real    0m23.989s
user    0m11.405s
sys     0m4.984s

Notice the 1sec (17%) less system time. This is on an
otherwise completely idle system, so the spinlock bouncing
between CPUs should be non-existent. I would expect the
effect on a real system to be higher. The 17% are not
entirely repeatable, but the mutex code was at least 10%
faster than the fcntl code measured in system time.

Volker

-- 
SerNet GmbH, Bahnhofsallee 1b, 37081 Göttingen
phone: +49-551-370000-0, fax: +49-551-370000-9
AG Göttingen, HRB 2816, GF: Dr. Johannes Loxen
http://www.sernet.de, mailto:kontakt at sernet.de