[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <035101cca94d$2a0e54e0$7e2afea0$@net>
Date: Tue, 22 Nov 2011 13:30:16 -0600
From: "jfoug" <jfoug@....net>
To: <john-dev@...ts.openwall.com>
Subject: RE: SHA1 SSE2i R&D work
>Good stuff. What's the gain for dcc2?
>magnum
Still POC. I have only made changes in the easiest of the formats (raw-sha1). I have not started on any other format yet.
However, I would assume the improvement would be at least 1/3, likely more, since for dcc2, we have multiple buffers, both inputs and outputs, and this change significantly improves both cases. Also, for dcc2, we are in a tighter loop (much tighter), with almost no impact on the overall throughput, outside of that deepest inner loop. There is no charge for key setup, manipulation, etc, once we reach that inner loop. So much of the cost is memory access, along with the SSE2 usage costs. We have reduced our memory footprint from 80+7 O(87) to 16+7+1+16 O(40), so we are using less than 1/2 of the memory we were using before.
Jim.
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.
