Re: Benchmarking

At 2004-04-07T23:20:48Z, Ivan Voras <ivoras_at_fer.hr> writes:

> And what do you call the results of bonnie++ sequential IO output test (in
> the table)? :) This is what I was reffering to in the cited text.

I think I was cross-eyed at that point.  I'd just finished looking at the
filesystem throughput test.

>> You describe your custom FreeBSD 5 kernel as keeping the WITNESS options.

> Really? Where do I say that?

When you didn't say that you'd removed it.

> The RELEASE kernel of FreeBSD are without WITNESS, and my CUSTOM kernel
> was based on it. The options I gave were the differences between CUSTOM
> and GENERIC.

You don't have "WITNESS" and "WITNESS_SKIPSPIN"?  I didn't realize that they
were temporarily removed from the -RELEASE branches.

> I admit NetBSD was sort of an afterthought, but do you really mean that
> the physical position and of the /bench slice would affect the tests in
> significant ways?

Yes.  The physical position of the /bench slice could have a huge effect on
performance at the hardware level.  Also, changes in the size of the disk
affect the number of created inodes, free working space (remember that
filling the last several percent of a filesystem has a drastic performance
penalty), and all sorts of other thigs.

> Even if the bonnie++ results are completely invalid, other results remain
> valid.

Except where the above changes have a large impact on IO performance.

Also, I'd really like to see your tests repeated with SCHED_4BSD.  Although
most people agree that _ULE is the way of the future, there have been some
grumbles of less-than-ideal behavior as of yet.  It's a new system, and it
may have a few bugs to iron out.

>> Basically, you ran some tests on divergent systems and got some results,
>> but that's about the only conclusion I was able to get from it.

> You're perfectly entitled to :)

I don't mean to make light of what you've done.  After all, you don't see
*my* benchmark page lying around, do you?  :)  But if you get your wish and
this is mentioned on Slashdot, then expect a whole load of gripe to come
your way.  I'm at least trying to be jovial about it.  :)
-- 
Kirk Strauser

"94 outdated ports on the box,
 94 outdated ports.
 Portupgrade one, an hour 'til done,
 82 outdated ports on the box."