[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAKtfLcttndkq0h1Ui1QgesMRTvv_38BnBd7yAQVi-2xL0Mvqtw@mail.gmail.com>
Date: Sat, 13 Oct 2012 16:52:56 -0400
From: Alain Espinosa <alainesp@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: bitslice DES on GPU
On 10/13/12, Sayantan Datta <std2048@...il.com> wrote:
> As a point of reference, what should be our targeted non-overhead speed?
> For instance, Hashcat does 83.4M c/s in traditional-des.
Hi Sayantan. I am implementing bitslice LM algorithm in GPU for the
next version of Hash Suite. In an AMD HD 6450 i am benchmarking 80M
when loading a lot of hashes(1000-10000) and 250M-300M for small
hashes (less than 32) (oclhashcat get 24M). Yet there is room for
improvements, the ALUBusy is ~60%-70%. So i can probably double the
speed for small hashes, and get 3x-4x speedup for a lot of hashes.
In my opinion the target speed of a bitslice LM GPU implementation is
near Hashcat NTLM speed.
saludos,
alain
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.
