On Tue, Nov 03, 2009 at 07:09:57PM +0100, Karsten Nohl wrote: > > On Nov 3, 2009, at 6:48 PM, sascha wrote: > > > On Tue, Nov 03, 2009 at 06:12:00PM +0100, Karsten Nohl wrote: > >> We use a mix of these two techniques but stay closer to the > >> distinguished points scenario. > >> Parameters for the tables being computed at the moment are: > >> M = 64 (A5/1) > >> D = 38 (4 TB) > >> C = 6 (50% success rate) > >> => T = 33.5 (2 minutes on a computer with GTX260 that has all tables; > >> less when the distributed cracking network is used) > >> > > This number is not correct. For a lookup of a single value, you have > > to compute > > (1 + 2 + 3 + ... + 32) * 2^15 A5/1 rounds, which is 528 * 2^15 = > > 17.3 million > > To do the lookup in all 400 tables, you need 7 billion rounds. > > 7 billion links can be computed in 42 seconds on a GTX260, and you > > have to do this for 204 keystream samples, totaling 8500 seconds on > > a single node with all the tables and 21 seconds with 400 nodes > > sharing > > the computation. > > that's correct -- my bad. I was using the formula for pure > distinguished points tables.
And i guess with all tables having no round function which is impossible to do because of the merges. To show rationale for rainbow tables again, with 400 * 32 DP tables (having the same number of merges as 400 rainbow tables), you have to compute 32 * 2^20 * 400 * 204 A5/1 rounds instead of 528 * 2^15 * 400 * 204, which is half as much. _______________________________________________ A51 mailing list [email protected] http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51
