[viff-devel] Secure Multi-party AES

[Martin Geisler]

Marcel Keller <mkeller at cs.au.dk> writes:

Hi Marcel

> I have now written some thoughts on the AES implementation in
> VIFF. Interestingly, my analysis of the computing time of the
> different variants (regarding the inversion) contradicts to the
> observed computing time. Feel free to find the error(s).

Did the timings behave better when you started doing more AES
operations in parallel?

As we discussed at a meeting here at DAIMI, I usually run 1000
multiplications in parallel in order to minimize the impact of the
network latency on each multiplication.

-- 
Martin Geisler