Hi everybody, Ivan told me how we can implement pseudo-random zero-sharing over a degree 2t polynomial. It even uses most of the stuff we already have so I went ahead and implemented it.
I then make a prss_generate_triple method which uses PRSS-based methods instead of the single_ and double_share_random methods (they had all the hyper-invertible matrix stuff going on for validation). Strangely the time for preprocessing has not improved... It stayed at an average time of about *20 ms* for a multiplication triple both before and after the change -- I don't understand that :-( At first I had only replaced the double_share_random method (but kept the single_share_random) and there the time for preprocessing did go down from about 20 ms to 12 ms. I'm thinking that there might be some unfortunate overhead in the preprocessing book-keeping. We should try running benchmark.py under a profiler to see where the time is spent. -- Martin Geisler _______________________________________________ viff-devel mailing list (http://viff.dk/) viff-devel@viff.dk http://lists.viff.dk/listinfo.cgi/viff-devel-viff.dk
