Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Date: Mon, 25 May 2015 22:51:48 +0200
From: Lukas Odzioba <lukas.odzioba@...il.com>
To: john-dev@...ts.openwall.com
Subject: PHC candidates project - irc minutes

This thread is ment to gather notes from irc meetings - currently mine
and Agnieszka's but I believe solar will be involved too. Just the
major points of the discussion are published here. I hope this will
help in co-mentoring.

--------------------------------------#1
21.05.2015-------------------------------------

A: Found problem with self test on parallel-opencl, generated
(LWS,GWS) settings are not optimal.

L: Reviewed code and pointed potential root cause of the problem in
self-test parameters. Suggested moving discussion on john-dev, and
trying to implement splitted kernel.

--------------------------------------#2
23.05.2015-------------------------------------

L: Suggested viewing atom’s “Optimizing the Computation of Hash
Algorithms as an Attacker”

A: Agreed to do it.

--------------------------------------#3
24.05.2015-------------------------------------

L: Suggested making project’s wiki page to document project advances.

A: Agreed to do wiki page.

L: Sent link to OpenCL books (http://bookfi.org/s/?q=opencl&t=0) and
suggested focus on chapters about optimizations to find new ideas.

--------------------------------------#4
25.05.2015------------------------------------

A: Looking for suggestions about possible improvements of parallel-opencl.

L: Reviewed current parallel-opencl implementation and parallel pdf spec.

A: Considered making interleaved and later vectorized implementation.

L: Suggested simplifying the current one before moving to such tasks.

A: Suggested moving parallel loop to the GPU and keep the rest on cpu side.

L: Said let’s keep it simple for now and reconsider later.

L: Suggested getting rid of general INIT UPDATE FINAL SHA512 and
making an optimized version that assumes some input bytes be always 0
- no need to initialize them and also some result of some operations
is known.

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.