On Wed, 19 Mar 2003, Benjamin Vander Jagt wrote:

> I haven't done any hard benchmarking yet.  Here's what I have now.  I'm
> in RH8.0 right now, and playing an MP3 in XMMS took 0.2% of my CPU with
> the precompiled i386 installation and the MP3 plugin RPM from GURU
> labs.  Simply rebuilding everything with the --target=athlon (nothing
> else), I play MP3s with 0% CPU utilization.  I'd have to do some long
> term statistical logging to give you firmer numbers.

Now yer attempt at any sort of useful numeric benchmark is noble..but this
is particularly not useful. Why, you may ask? Well....

1) a change from 0.2% to 0% is a measure of infinite improvement...find a
better way to get a measureable benchmark. And even so I'm not even sure
what you are measuring with this. ps's man page remarks on %CPU with:
"It is time used divided by the time  the  process has been running. It
will not add up to 100% unless you are lucky"

how is this a useful benchmark? Can't a number of other factors..like
other running processes get in the way of this measurement?  This only
becomes significant with reasonable statistics, and careful attention to
the other things in the system are doing. But I would find a better
measurement than this. When the man page sez you cant expect to see the
number add up to 100%, a change of 0.2% has got to be in the accepted
noise of the system, and invalidates the measurement....
this is rule #3 of junk science.
http://chronicle.com/free/v49/i21/21b02001.htm

2) Your using 3rd party rpms/srpms....GOD ONLY KNOWS what settings were
used to compile that original binary rpm from GURU labs. There are several
levels of optimization. Did the guru labs compile that binary with
-mcpu=i686? which is different than specifying -march=i686. Did GURU use
-O2? AFAIK, redhat compiles their binaries for x86 with -mcpu=i686 and
-O2.  Go back....recompile the srpm for i386 with the Red Hat's default
optimization settings...then recompile with yer -march change.  If yer
going to pretend to make a benchmark...make sure you know what the
original binary was compiled to use, before you pretend to make a
comparison.

> CPU optimizations can make a huge difference.  Consider how MMX and
> 3D-Now! greatly reduce the program size, (and therefore) memory
> consumption, and processor usage.

Look at all the things in that sentence you talk about..and the only thing
you include is the dubious %CPU stat.  Do it right...or don't do it.


-jef"trained to assume data is wrong..because his data usually is"spaleta



-- 
Phoebe-list mailing list
[EMAIL PROTECTED]
https://listman.redhat.com/mailman/listinfo/phoebe-list

Reply via email to