[viff-devel] Faster two player runtime

[Martin Geisler]

Good evening everybody,

I have incorporated the changes suggested by Claudio here

  http://article.gmane.org/gmane.comp.cryptography.viff.devel/304

and I am now able to run a multiplication in 585 ms using two of the
DAIMI thyra machines. The changes have improved the running time by a
factor of six!

The protocol is asymmetric, and part of this speedup comes from making
the two players alternate with each operation so that they end up
doing an equal amount of work on average. Without that I measured
times around 900 ms per multiplication.

With some unit tests and documentation I think we are ready to include
this code in VIFF.

-- 
Martin Geisler