Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Mon, 17 Aug 2015 12:53:39 +0300
From: Solar Designer <>
Subject: Re: Formats using non-SIMD SHA2 implementations

On Mon, Aug 17, 2015 at 03:50:11PM +0800, Lei Zhang wrote:
> BTW, how do I define ALGORITHM_NAME for this kind of format where different algorithms are involved? What about "AES 32/64 SHA256 256/256 AVX2 8x"?

We haven't been separating the bitness per crypto primitive so far.
I think that if most processing time is spent in SHA-256 and very little
in AES, we should only list the SHA-256 code's bitness here.

If we have a format where substantial portions of processing time are
spent in different bitness implementations of different primitives, then
we have something to decide upon here.  Probably it still makes sense to
report just the maximum, but Lei's suggested syntax also makes sense.

BTW, the 32/64 for AES might not be right.  Not when we're using AES-NI,
which operates on 128-bit SIMD registers but not treating them as SIMD.
I think 32/64 was previously reported for SHA-256 only (since that's
what most processing time was spent on?), not for AES.


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.