[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <551DB7DB.7000905@dei.uc.pt>
Date: Thu, 02 Apr 2015 22:42:51 +0100
From: Samuel Neves <sneves@....uc.pt>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] Compute time hardness (pwxform,blake,blamka)
On 02-04-2015 22:22, Marcos Simplicio wrote:
>> BTW, you can take advantage of the (a+b)+2ab latency optimization in
>> > your defensive code as well. It might cost an extra register and a move
>> > instruction, though, so it isn't universally an improvement. It might
>> > improve the latency on new CPUs (Ivy Bridge and Bulldozer, and newer)
>> > and make it worse on older CPUs (Sandy Bridge and older when running
>> > non-AVX builds).
> That seems worth trying (actually, we need to find someone with better
> implementation skills than ourselves: most of Lyra2's code tricks come
> from Blake2b's original implementation...). Thanks!
>
We used this trick (replacing + with ^, * with &) in NORX (pg. 23 of https://norx.io/data/norx.pdf); it does noticeably
improve latency.
Powered by blists - more mailing lists