Subject: Re: Multiprocessor with NetBSD ?
To: NetBSD-current Discussion List <>
From: Greg A. Woods <>
List: current-users
Date: 06/06/2001 16:21:09
[ On Wednesday, June 6, 2001 at 14:24:20 (-0500), mike stone wrote: ]
> Subject: Re: Multiprocessor with NetBSD ?
> my last post was a weak attempt to compare the relative merits of
> AP -v- SMP.   while SMP does have its strengths, it also has weaknesses.
> and like any technology, SMP is best suited to a particular problem set.
> AP also has strengths and weaknesses, and a range of problems for which
> it happens to be useful.

Oh, indeed!  I agree very much!

> my question is "how well does SMP support the work we're most likely
> to do?" with the corrolary, "would AP provide equal, better, or worse
> support for that same work?"

I think that given the most commonly available multi-processor hardware
available today that building AP for NetBSD would be a step backwards.

I don't have an exact reference to cite, but it would seem to me that a
full SMP implementation of NetBSD will give the best overall general
purpose computing performance for a Unix environment.  As far as I can
remember this is the same direction every commercial Unix vendor has
headed in and that kind of consensus just doesn't come from market
pressure alone.

I have another interesting text book on the subject:  "High-Performance
Computer Architecture" by Harold S. Stone (AW, 1990) of which about 1/3
of the content focuses directly on multiprocessing (and perhaps another
1/3 describing closely related topics).  The author presents several
models for multiprocessor architectures and does some basic mathematical
analysis of each to show their strengths and weaknesses.  Unfortunately
I don't know enough about modern (or even semi-modern) commercial
multiprocessor implementations to know how their architectures fit onto
his models.

(BTW, my remark about the "von Neumann bottleneck" wasn't directed
specifically at you -- it was just a general comment to everyone reading
the thread, based on what I was reading last night in "Computer

							Greg A. Woods

+1 416 218-0098      VE3TCP      <>     <>
Planix, Inc. <>;   Secrets of the Weird <>