Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!wuarchive!gem.mps.ohio-state.edu!apple!oliveb!pyramid!csg
From: csg@pyramid.pyramid.com (Carl S. Gutekunst)
Newsgroups: comp.sys.sequent
Subject: Re: Dynix licensing
Keywords: license, user-limits
Message-ID: <86270@pyramid.pyramid.com>
Date: 2 Oct 89 21:16:54 GMT
References: <6006@wolfen.cc.uow.oz> <294@baird.cs.strath.ac.uk> <12288@boulder.Colorado.EDU>
Reply-To: csg@pyramid.pyramid.com (Carl S. Gutekunst)
Organization: Pyramid Technology Corp., Mountain View, CA
Lines: 20

>I don't purport to speak for Sequent, but if I were in their shoes, I'd
>keep tight control on the parallel kernel sources.

Actually, *any* kernel sources. There is a lot of value-added that can be
done to the UNIX kernel. Stright multiprocessing isn't very special any more
(and for that matter, Dynix isn't doing anything particularly unusual). But
there is a lot of closely-guarded stuff regarding scheduling algorithms (for
CPU, I/O, and Disk), processor affinity, distribution of interrupts, costs of
context switches and cache/TLB flushes, and so on. Even SVR4 is a long way
from what I'd call a "commercial" UNIX; you have to add a lot to it to make
a production system. *That* is what gives companies their edge, and is why
most companies are so protective of their sources.

>it's my opinion that Sequent's kernel is the cleanest/most elegant/best
>multiprocessor Unix kernel on the market.

So how many *other* MP kernels have you seen? I suspect the UMAX and MACH
people would love to discuss this with you.