Re: SMP scalability: 8 -> 32 CPUs

William J. Earl (wje@fir.engr.sgi.com)
Mon, 30 Nov 1998 23:49:24 -0800

Henrik Olsen writes:
> On Mon, 30 Nov 1998, Stephen C. Tweedie wrote:
...
> > "Let's assume that we invent some mechanism for binding 1024
> > cpus into a single box. I know it doesn't exist yet, but how
> > fast will it be?"
> > Binding that many CPUs together is hard! You tell me how you plan to do
> > it, and _then_ I'll tell you whether clustering is faster. :)
> > --Stephen
>
> There is a very simple answer to this:)
> Unless it's a problem that will take years to complete, the cluster will
> finish the job sooner (only metric that really makes sense), since you can
> make one now.
...

An SGI Origin 2000 can in principle support up to 1024 processors
in a single system, with relatively low memory latency between the
most distant processor and memory pairs. We have two 256 processor
systems running and many 128 processor systems shipped. Scaling an
operating system for general purpose use on such a system is of couse
not simple (which is why we don't have any 1024 processor systems).
The basic hardware components, however, have been shipping for some
time, so the "mechanism for binding 1024 cpus into a single box" does
exist.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/