Re: [g.mcgarry%ieee.org@localhost: More Microbenchmarks]

[Thor Lancelot Simon <tls%rek.tjls.com@localhost>]

On Tue, Jun 03, 2008 at 03:27:28PM +0200, Johnny Billquist wrote:
> I'm curious. Why would it be of historical interest only, if we can 
> produce numbers that show that the system worked much more efficiently 
> in the past? (Ie. getting services done took less real time, using the 
> same hardware, with similar loads.)

I'm not interested in using benchmark numbers to grind my axe.  If you
actually look at what Gregory/Andrew posted, to which I responded, you'll
see that some parts of the system got faster, and some parts got slower.
From macrobenchmarks (e.g. the mysql runs Andy does) and from Sun's
microbenchmarks we know that for many important workloads, as well as in
many basic ways lmbench is too old or insufficiently comprehensive to
measure, the system is, in fact, much faster than it used to be.

You can't tune a system for every workload -- or every hardware it might
ever run on -- at once.  Given that, trying to resuscitate my decade-old
system that I benchmarked 1.2 on seems like a significant waste of time.

Investigating the parts of Gregory's results that make me scratch my head
and say, "huh, that's strange" -- that's of more interest to me, at least.

Thor