Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 11 Mar 2015 22:45:19 +0100
From: magnum <>
Subject: Re: bitslice MD*/SHA*, AVX2

On 2015-03-11 22:21, Solar Designer wrote:
> On Wed, Mar 11, 2015 at 11:55:09PM +0300, Solar Designer wrote:
>> There's definitely some further room for optimization in md5slice.c -
>> for example, it does not yet take advantage of the common subexpressions
>> in the invocations of HH() (we can save 8*32 XORs).
> BTW, it appears that we don't currently use this optimization.  Not even
> in the copy of my portable MD5 implementation in JtR, which I've since
> updated to include this optimization here:
> In my testing, this might not be beneficial on 2-operand archs such as
> plain x86, but it should be on 3-operand archs such as AVX.  So we
> should update the code in sse-intrinsics.c, and benchmark.  And we should
> update the plain C code anyway, such as for non-x86 archs (which are
> mostly 3-operand RISC).
> magnum, Jim?

Yeah... unless we have some GSoC candidate wanting to show his/her
teeth? That would be a good start!


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.