The reason that the second, third, fourth etc lookup is faster is that all the 16320 required table lookups are already cached in memory. The first time, it needs to read it from disk. If you get faster storage (e.g SSD-disks), you will get much better results.
-- Best Regards, Joel Eriksson On Fri, Jun 17, 2011 at 2:50 AM, hunting <hunt...@foxmail.com> wrote: > Dear all: > > Recently, we do some experiments with the berlin tables and the kraken > code. The two graphic cards we use are of the type ATI HD 5670. To store the > rainbow table, the traditional HDDS type disk is adopted. We have done > several times of experiments. However, a phenomena which is described below > always appears and it semms so head-scratching. > > Each time we crack a new 114-bit stream, the time needed for table > looksup is always aobut three minutes, while for the same bit stream to be > cracked in the second, third, fourth, ... time, the task can be > finshed within 10 seconds (about 7~9 seconds) . Thus, when we want to crack > a whole 456-bit stream, we will consume about 12 minutes which is > time-consuming and is inefficient for most applications. > > Thus, I wonder why it always needs three minutes for the first time to > crack the 114-bit stream. Can this time be significantly reduced, for > example about 10 seconds just as the second or third trial? > > Wish all your help. > > Regards. > Lanma > _______________________________________________ > A51 mailing list > A51@lists.reflextor.com > http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51 > > -- Best Regards, Joel Eriksson _______________________________________________ A51 mailing list A51@lists.reflextor.com http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51
