[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20121209235201.GB4421@openwall.com>
Date: Mon, 10 Dec 2012 03:52:01 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: bitslice DES on GPU
On Sun, Dec 09, 2012 at 02:11:56PM +0530, Sayantan Datta wrote:
> I checked the tahiti isa of two different hardcoded kernels and to my
> surprise they just differ by a single line.
> v_xor_b32 v68, v38, v68 // 00000A24:
> 3A888926
> vs
> v_xor_b32 v68, v21, v68 // 00000A24:
> 3A888915
It is unreasonable to comment on compiled code differences without
having seen your corresponding source code. Maybe it's such that your
hard-coded constants got optimized out (in both cases) as not affecting
the result. It is possible that there are some other left-over
differences from the optimizer (such as in register numbers used).
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.
