Subject: Re: sysbench and ubench benchmarks
To: None <firstname.lastname@example.org>
From: Blair Sadewitz <email@example.com>
Date: 05/01/2007 06:27:10
Yes, I just can't grok how the NexentaOS memory score could possibly
be accurate. Also, I think it's important to benchmark UP v. MP
machines to discern differences in scheduler implementations.
Many people looking to run a large MySQL server, for example, will be
using a multi-processor machine.
Then there's also the issue of statistical significance (p < 0.01),
which these benchmark runs completely ignore. This isn't a matter of
"correct" or "incorrect" results per se, but rather, as others have
said, the degree to which one should take these benchmarks as an
indicator of anything.
Moreover, OpenBSD is not *that* difficult to get installed, unless
there was some hardware-specific quirk that genuinely precluded
Then, of course, there's the fact that "brute speed" isn't necessarily
the most important factor in OS choice!