[Beowulf] How to configure a cluster network

[Peter St. John]

I imagine a hybrid topology of certain sized subclusters connected
internally with a right topology for their size, and the subclusters
connected to each other with some other topology, etc. The way cores on a
chip are connected is different obviously from the way chips on a board, or
boards on a backplane, or boxes to routers, or routers to metarouters...and
I need hybrid-ness so that the optimization/self-reconfiguring algorithm can
do something with its platform.

So maybe just some scale component or subcluster would be hypercube, some
FNN, some tree, I don't know. I want enough wires with enough nodes so that
my application can tell **me** what topology works best for a compuational
category of applications.

Peter

On 7/25/08, Lawrence Stewart <larry.stewart at sicortex.com> wrote:
>
> Mark Hahn wrote:
> >> It is my sacred duty to rescue hypercube topology. Cool Preceeds
> >> Coolant :-)
> >
> > I agree HC's are cool, but I think they fit only a narrow ecosystem:
> > where you don't mind lots of potentially long wires, since higher
> > dimensional
> > fabrics are kind of messy in our low-dimensional universe.  also, HC's
> > assume intelligent routing on the vertices, so you've got to make the
> > routing overhead low relative to the physical hop latency.
> >
> > it does seem like there is some convergence to using rings onchip,
> > fully connected graphs within a node and fat trees inter-node.
> > one unifying factor is that these are all point-to-point topologies...
>
> Hypercubes give log diameter, which is good, but when you grow the machine
> you have to add more ports to each node, which is not so good once you run
> out of pins.
>
> Other topologies, such as Kautz and deBruijn graphs, give log diameter
> as well,
> but with a fixed number of ports per node, so you can put the node on
> one chip
> and still build systems of greater or lesser size without having to respin.
>
> You can route arbitrary size Kautz graphs on a
> fixed number of layers, so when these somewhat wacky topologies
> go on chip you can still route it on N metal layers.
>
> The point about long wires is well taken, but I think it is the price of
> low diameter.
>
>
> --
> -Larry / Sector IX
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.scyld.com/pipermail/beowulf/attachments/20080725/3d4f4ccd/attachment.html