crypt-dev - Re: alternative approach

Follow @Openwall on Twitter for new release announcements and other news

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20110614013407.GB25262@openwall.com>
Date: Tue, 14 Jun 2011 05:34:07 +0400
From: Solar Designer <solar@...nwall.com>
To: crypt-dev@...ts.openwall.com
Subject: Re: alternative approach

Yuri -

On Mon, Jun 13, 2011 at 09:59:26PM -0300, Yuri Gonzaga wrote:
> The first code I sent has wrong results (131 LUTs) because I did
> modifications on the code after simulating.

In other words, you synthesized, tested for correct operation, then made
incorrect changes/optimizations, synthesized again, did not test,
reported the results for the latter revision of the code (non-working).
Is this the correct understanding?

> So, next time I worked in this project, I realized the modifications set it
> not to work properly.
> Then, I corrected this and sent a new code to the mailing list.
> After the correction, the results become different (938 LUTs).
> And this is the current status. The last pdf has results related to the
> correct code.

OK.  That PDF shows the smallest LUT count for the full pcadd(), as
opposed to the various alternatives.  This is inconsistent with your
earlier statement that you got reduced LUT count when you tried
replacing pcadd() with the shorter expressions.  Which is right?

Here's what I think we should start doing:

Each time you report synthesis results, please provide two files: one
file with full source code (something-someversion.v) and the other with
synthesis results (something-someversion.html or .pdf or whatever).
If you need to provide 4 results - such as for different pcadd()'s -
provide a total of 8 files.

This involves some overhead, but our approach so far is obviously not
reliable enough.

And please use our wiki for file uploads.  For example, create this page:

http://openwall.info/wiki/crypt-dev/files

Then your uploaded files will fall under crypt-dev (DokuWiki places them
one level above the page being edited).

How does this sound to you?

Thanks,

Alexander

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.