Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [day] [month] [year] [list]
Date: Mon, 8 Oct 2012 22:26:15 +0530
From: Dhiru Kholia <dhiru.kholia@...il.com>
To: john-dev@...ts.openwall.com
Subject: Keccak Patch

Hi,

I have used official optimized implementation as base. This code is just a PoC.

$ ../run/john -fo:keccak -t # FX-8120 with XOP, 1 core
Benchmarking: Keccak 512, E = 64 [32/64]... DONE
Raw:	2029K c/s real, 2029K c/s virtual

$ ../run/john -fo:keccak -t  # FX-8120 with XOP, 8 cores
Benchmarking: Keccak 512, E = 64 [32/64]... (8xOMP) DONE
Raw:	6163K c/s real, 772362 c/s virtual

In comparison, hashcat does 3.28M on one core and 12.3M using 8 cores.
Lot of performance loss can be attributed to OpenMP.

Results without XOP aren't too bad.

$ ../run/john -fo:keccak -t # FX-8120 without XOP, 8 cores
Benchmarking: Keccak 512, E = 64 [32/64]... (8xOMP) DONE
Raw:	5036K c/s real, 631081 c/s virtual

How do I make sure that XOP instruction set is being used?

This patch compiled easily on one machine but on bull machine I had to
disable DES (family) formats.

-- 
Cheers,
Dhiru

Download attachment "0001-Keccak-PoC.patch.gz" of type "application/x-gzip" (20458 bytes)

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.