Date: Wed, 18 Sep 2013 20:16:59 +0200 From: Katja Malvoni <kmalvoni@...il.com> To: john-dev@...ts.openwall.com Subject: Re: bcrypt-parallella on 64-core (was: Katja's weekly report #13) Hi Yaniv, These are register values from cores which reads start flag correctly and reaches idle instruction: [0x000f0400] = 0x00080000 [0x000f0404] = 0x00001000 [0x000f0408] = 0x000007e8 [0x000f040c] = 0x00000000 [0x000f0410] = 0x00000000 [0x000f0414] = 0x00000000 [0x000f0418] = 0x00000000 [0x000f041c] = 0x00000000 [0x000f0420] = 0x00000858 [0x000f0424] = 0x000003ff [0x000f0428] = 0x00000000 [0x000f042c] = 0x00000000 [0x000f0430] = 0x00000000 [0x000f0434] = 0x00000000 [0x000f0438] = 0x00000000 [0x000f043c] = 0x00000000 [0x000f0440] = 0x00000000 [0x000f0444] = 0x00002600 [0x000f0448] = 0x00000000 But cores which don't read and update start flags have these values: core (3, 5) has zero on all addresses except [0x000f0444] = 0x00003500 core (4, 6) has zero on all locations core (7, 4) has zero on all addresses except [0x000f0444] = 0x00000200 core (7, 5) has zero on all addresses except [0x000f0444] = 0x00002600 It seems that some cores don't start computation. So far it looks random, only regularity I see is that first two rows work (If I use only first two rows I get 1207 c/s). Katja Content of type "text/html" skipped
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.