[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20120215160206.GA2613@openwall.com>
Date: Wed, 15 Feb 2012 20:02:06 +0400
From: Solar Designer <solar@...nwall.com>
To: john-users@...ts.openwall.com
Subject: Re: OMP version not use all CPU
On Wed, Feb 15, 2012 at 07:46:22AM -0800, Alain Espinosa wrote:
> On 2/15/12, Solar Designer <solar@...nwall.com> wrote:
> > What c/s rates are you getting for these tests, though? I am interested
> > in both --test and actual runs.
>
> Option --test was tested on the same Core i3 machine mentioned in
> "http://hashsuite.openwall.net/performance".
Yes, but my question was about speeds of john-omp.exe - I don't think
you provided any numbers for that (except for CPU time use percentages).
> I give a try to NT2
> format also ("real" and --test), and found it a very little speed-up
> compared with NT. I assume that using 64 bits john will be different.
> Some numbers for 1,10,100...(same as webpage) hashes:
>
> John-1.7.9-j5[nt2] 19.41 18.9538 18.51 16.544 16.5446 12.019226 11.16 5.122
>
> I will add all those numbers to the performance webpage mentioned earlier.
OK. There was in fact a known slight performance regression with NT in
1.7.9-jumbo-5; it's been addressed in magnum's tree, so the next jumbo
should have faster NT (should be about the same as NT2 then).
And perhaps you can contribute some optimizations, too. ;-)
Alexander
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.
