The patch delivered 1.45x for SHA256 and 1.3x for SHA512, compared to
OpenSSL 1.0.1 (current at that time). At the very least, it showed (for the
first time) to OpenSSL developers that much higher performance was
achievable. The purpose of the patch was to demonstrate the benefit of the
n-SMS method, therefore no additional optimizations were made to the
OpenSSL “body” code itself.

The response to the patch submission could have better expressed some
credit for that.



Furthermore, the implementation in
http://git.openssl.org/gitweb/?p=openssl.git;a=commitdiff;h=c4558efbf3a44a1b5e68dce46347dd3888db4760
is
using 2-SMS for the AVX2 version. However, it forgets the due reference to
http://rt.openssl.org/Ticket/Display.html?id=2784&user=guest&pass=guest(and
by now, also: S. Gueron, V.Krasnov, “Parallelizing message schedules
to accelerate the computations of hash functions”,

Journal of Cryptographic Engineering November 2012, Volume 2, Issue 4, pp
241-253).

This paper was about n-SMS where n is the number of data blocks whose
“message schedule” is processed in parallel using SIMD instructions. It is
based on a general observation on Davies-Meyer hashes, and specifically,
the values n=2, 3, 4 are mentioned as options there. The paper gave an **
example** of 4-SMS and 8-SMS, and the patch executed 4-SMS for SHA256,
which we thought was a good choice.



Using 2-SMS, mixed with some other techniques (BTW, due to some other
author) does not exempt from proper citing in the Git version that would
enter the library.



Regards, Shay Gueron

**************************

Professor Shay Gueron

Department of Mathematics

University of Haifa

Haifa 31905  ISRAEL

s...@math.haifa.ac.il

**************************








-----Original Message-----
From: Andy Polyakov via RT [mailto:r...@openssl.org]
Sent: Saturday, March 30, 2013 12:11
To: Gueron, Shay
Subject: [openssl.org #2784] Resolved: [PATCH] Eefficient implementations
of SHA256 and SHA512, using the Simultaneous Message Scheduling method

According to our records, your request has been resolved. If you have any
further questions or concerns, please respond to this message.
---------------------------------------------------------------------

______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
Development Mailing List                       openssl-dev@openssl.org
Automated List Manager                           majord...@openssl.org

Reply via email to