[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 18 Dec 2011 20:18:15 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: 1.7.9-jumbo
On Sun, Dec 18, 2011 at 05:03:50PM +0100, magnum wrote:
> The change is that in 1.7.8, we always had 16K keys_per_crypt, and in
> 1.7.9 we have 8K x number of threads (the latter is capped to 4, because
> it does not scale further). I did that on some AMD hardware using
> gcc-4.5 and I did not see this performance drop (on the contrary).
The code in 1.7.8-jumbo-8 did scale beyond 4 threads on 2xE5420:
Benchmarking: CRC-32 [32/64]... (8xOMP) DONE
Many salts: 250200K c/s real, 31275K c/s virtual
Only one salt: 26133K c/s real, 3295K c/s virtual
Benchmarking: CRC-32 [32/64]... (8xOMP) DONE
Many salts: 250003K c/s real, 31211K c/s virtual
Only one salt: 26411K c/s real, 3293K c/s virtual
Benchmarking: CRC-32 [32/64]... (4xOMP) DONE
Many salts: 151187K c/s real, 38174K c/s virtual
Only one salt: 22200K c/s real, 5536K c/s virtual
Benchmarking: CRC-32 [32/64]... (4xOMP) DONE
Many salts: 161726K c/s real, 40431K c/s virtual
Only one salt: 39976K c/s real, 9969K c/s virtual
(same "john" binary, I am using OMP_NUM_THREADS=4 on the last two runs).
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.
