> Re: Pelica+ vs. Mustang:  it would be interesting to see a comparison
> of these (perhaps against a similar generation box such as a sandpiper)
> using lmbench, which is available at:

I use lmbench all the time.  Have been for a long time, actually.

However, i've explicitly avoided posting results for NetBSD/Alpha
obtained with it for a two reasons:

	(1) microbenchmarks are not necessarily indicative of overall
	    system performance, and in my opinion people put too much
	    emphasis on them.  (i.e. "principle")

	(2) I've made absolutely no attempt to optimize large parts of
	    the NetBSD/Alpha kernel, simply because i don't have time,
	    and I know (from those microbenchmarks and/or profiling
	    8-) that at least some parts of NetBSD/Alpha are grossly
	    inefficient, or at least not as good as they could be.

I have no objections to benchmarking NetBSD/Alpha on multiple machines
and posting 'relative performance,' but i don't have enough machines
to make it interesting.


chris