On Wed, 2 Feb 2000, Stefan Struiker wrote:
> Team M:
>
> Does the L-L process lend itself to some parallelization
> (not the running of multiple copies) over N processors?
>
> I have a short monograph on Amdahl's Law to squeeze out,
> and I thought I might use L-L as an example, bad or good.
Since the L-L test for 2^P-1 consists of a sequence of P-2 steps, each
consisting of squaring the result of the previous step and subtracting 2
(mod 2^P-1), the ability to parallelize is essentially lost, except for
possible multiprocessor implementations of the squaring step, which in
that case will be bound by the speed of inter-processor communication.
>
> Best To All,
> Stefanovic
>
> PS: Amdahl's Law relates Maximum Speedup to a fraction f ( < 1 ) of
> ops that must be done sequentially, and the number p of available
> processors:
>
> 1 / MaxSpeedRoom = f + (1 - f) / p
>
>
> _________________________________________________________________
> Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
> Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
>
--
Henrik Olsen, Dawn Solutions I/S URL=http://www.iaeste.dk/~henrik/
Darth Vader: Luke, come to the dark side.
Luke: No.
Darth Vader: Your goodness has redeemed me. Die, emperor scum.
Return of the Jedi, the Movie-A-Minute version
_________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
