[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <BLU0-SMTP157BE219E8508C8F104B2F6FDE10@phx.gbl>
Date: Mon, 25 Jun 2012 18:27:19 +0200
From: Frank Dittrich <frank_dittrich@...mail.com>
To: john-dev@...ts.openwall.com
Subject: Re: committed patches
On 06/25/2012 06:09 PM, magnum wrote:
> For now, I did not apply the Markov-split-by-percent. I personally think
> it's too near release and there may be dragons: Apart from the merge
> problems it will give me (albeit on non-public trees), what happens if
> you run MPI and use this option? And for that matter, what *should*
> happen if you do?
I just checked the source.
The split of the start:end range over MPI nodes happens
after all the other computations (convert % into real values, replace
end=0 with real end, ...) are finished.
I also made sure that the
fprintf(stderr, "End: %s converted to "LLd"\n", end_token, mkv_end);
happens only for mpi_id == 0 (if HAVE_MPI is defined).
So, MPI should work just fine.
Frank
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.
