cheap clusters

[Velocet]

On Wed, Apr 04, 2001 at 04:22:50PM -0400, Troy Baer's all...
> On Wed, 4 Apr 2001, Robert G. Brown wrote:
> > Speaking of DDR Athlons -- are there any motherboards that are actually
> > available out there?  Most of what I was able to find in a modest
> 
> There are a couple uniproc Athlon/DDR motherboards that are available
> right now if you look hard.  The most widely available ones are the
> 
> The major problem with DDR memory right now is finding DIMMs that are
> stable at the 133MHz system bus speed.  Some batches of Nanya PC2100-rated
> DDR RAM are very unstable (i.e. cause silent lockups) unless you downclock

Im lost here. Why are people buying the most expensive single
nodes for their computations? Arent we all building/running beowulfs
because no single machine could deliver the power we need? So why
get 10 super expensive DDR machines with super expensive DDR ram instead
of 40 less expensive machines? I dont understand the underlying
philosophy. Is the idea to get the most money in machinery into the
smallest space? Whats the criteria here? Is everything 'automatically'
considered 'cheap' because its not a $10M supercomputer here? (I remember
reading the word "cheap" in the Beowulf FAQ somewhere, but it doesnt
seem to be echoed by many people on the list, except a few who mainly
stay off the list and talk with me privately - for fear of being
chastized for the most cycles per dollar??)

There really needs to be a top500 GFLOPS/$ supercluster list. I bet
the current Top500 would be turned on its head. (even including cost
of power, a/c, network gear, cabling, cabinets, etc).

Mebbe Im wrong. Mebbe DDR ram isnt still 3 times more expensive in your
countries and the mainboards 2x as much as the average mainboard...
mebbe this is all cheap. To get something like 1.5 to even 2.5x
the performance for 2.5 to 3.0x the cost really doesnt seem to be
a win, and these are extremely conservative numbers. (I need to get
Atlas working with my copy of gaussian before I can compare directly.)

[ I realise there's a very fragile dynamic that can occur with
  parallel computations - a whole slew of slower nodes that actually
  adds up to more power (in Mhz/GFLOPs whatever) can be slower depending
  on network chatter and switch partitioning and many other factors, or
  the reverse is possible. Only detailed analysis of an actual cluster (or
  a very good similation) running the computations in question can give you
  the answers regarding what the best configuration here is. Obviously
  people are asking about equipment they have no access to test something
  this complex on, nor for extended periods in a reallife situation. Or perhaps
  there is a definite indication that there is economy in faster and faster
  single nodes for some types of computations and thats why everyone wants
  faster nodes NO MATTER the cost, the advantage is that great. I cant believe
  that thats true for all cases however. Can someone explain to me how this
  factor plays into the cost analysis? ]

Then again I dont have my cluster up yet (administrative details regarding
money) so I should probably shut my mouth and let the people who've built
clusters on a BUDGET talk. I'd be interested in hearing that it really is true
that 100 $500/node Tbird 800s beats the PANTS off 30 Pentium 4s for the same
price (the latter loaded with RDRAM). But I have no numbers, so I could be
quite wrong.

Very curious about it for sure, however. (Especially before I decide
to build my cluster out of mid range instead of bleeding edge parts! :)

/kc
-- 
Ken Chase, math at velocet.ca  *  Velocet Communications Inc.  *  Toronto, CANADA