[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAOLP8p55GPQM5CXh35Ebai-tUydP39SBf=W-vB6CjAQ7=juJMA@mail.gmail.com>
Date: Thu, 27 Mar 2014 21:09:33 -0400
From: Bill Cox <waywardgeek@...il.com>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] pufferfish
On Thu, Mar 27, 2014 at 8:55 PM, Jeremi Gosney <epixoip@...dshell.nl> wrote:
> On 3/27/2014 5:52 PM, Bill Cox wrote:
>> On Thu, Mar 27, 2014 at 8:22 PM, Solar Designer <solar@...nwall.com> wrote:
>>> I had actually considered revising the original bcrypt to introduce that
>>> "outer loop", and I felt that it could be done without it appearing as
>>> an extra loop. bcrypt already has its 2^cost loop - that one loop could
>>> be revised to move the 4 KiB region across a larger arena and to XOR (or
>>> whatever) it with random 4 KiB regions in the already-written portion of
>>> the arena.
>>>
>>> Alexander
>> +1 for this solution. That would make PufferFish competitive.
>>
>> Bill
>
>
> Implying it's not competitive...
Sorry... Even among geeks, I seem socially impaired.
I like PufferFish the way it is, but as Alexander pointed out, when
you bust out of L1 cache, performance drops. When you bust out of L3
into DRAM, if I read the code correctly, performance will drop a ton.
I'd be a hypocrite for not suggesting that you take his advice and
implement the wandering 4KiB window in the 2^cost loop. Over half of
my submission is built on ideas he suggested. I just hope there can
be some way for him to formally "win" regardless of which submission
built on his ideas wins.
I give myself credit for knowing good ideas when I see them. The
wandering 4KiB window seems like a winner to me.
Bill
Powered by blists - more mailing lists