Date: Sun, 2 Jul 2017 14:44:44 +0200 From: Solar Designer <solar@...nwall.com> To: john-users@...ts.openwall.com Cc: Denis Burykin <apingis@...nwall.net> Subject: Re: DES-based crypt(3) cracking on ZTEX 1.15y FPGA boards (descrypt-ztex) On Sat, Jul 01, 2017 at 02:08:36PM -0800, Royce Williams wrote: > Some quick keys_per_crypt tests on the cluster. > > A value of 2 was a ~5.4% improvement. Values above 2 were worse - > enough that I only let them run for a few minutes to verify. Thank you! It is unexpected to me that higher values slow things down not only in terms of passwords cracked by a certain time (which can be explained by the less optimal order), but also in terms of c/s. But that's what our testing is for. So maybe we should just make a multiplier of 2 the default. With one board, it increases the speed from 806M to 819M for me. (That's half way to the 831M figure I obtained with a multiplier of 8.) > Stock: > > 325g 0:00:59:59 0.49% (ETA: 2017-07-08 16:38) > 0.09029g/s 4550Kp/s 9881Mc/s 14078MC/s nujuaaaa..nujualks > > With keys_per_crypt *= 2: > > 369g 0:01:29:18 11.76% (ETA: 08:01:28) > 0.06886g/s 4585Kp/s 10454Mc/s 14840MC/s ayceaaaa 10454/9881 = 1.058 (or 5.8% speedup) 10454/14 = 747M per board (or about 91% of what I have with 1 board) Regarding the bugs you reported, I've just merged Denis' pull request today that I think should fix one of those. It also makes JtR wait and resume cracking if it's running with just 1 board and it experiences intermittent failure. I've just tested this latter property by extreme overclocking by 15% - it worked (the board got removed and re-detected during cracking, which resumed on its own). At 15% overclock the error rate is probably high enough that I am missing some crackable passwords, so it's not suitable for actual use. Per my testing, reasonable overclocks tend to be 5% to 10%. Alexander
Powered by blists - more mailing lists
Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.