[Beowulf] Good demo applications for small, slow cluster
Max R. Dechantsreiter
max at performancejones.com
Wed Aug 21 11:44:55 PDT 2013
> What about the old random number generator: take a 16 bit seed, square it,
> take the middle 16 bits, and repeat. They'd want a large number in order
> (so you can repeat an experiment, or a run of a model, with the same
> "random" numbers), and it's easy to computer sequentially; but if you want
> a million of them it would be nice to distribute the job, but I don't think
> you can. But maybe "can't parallelize" isn't the same as "badly inefficient
> to parallelize".
Many RNG _can_ be distributed. The enabling characteristic
is the existence of a way to "giant-step" the generator.
This is certainly possible for LCG (LCRNG), such as the one
used throughout NPB I.
The NPB generator has the form
x(k+1) = a*x(k) modulo 2**46 (suggested a = 5**13)
which has period 2**44.
It could be an interesting excercise: take two generators,
one which can be distributed, one which cannot....
More information about the Beowulf