[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20150328213827.GB14395@openwall.com>
Date: Sun, 29 Mar 2015 00:38:27 +0300
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: [GSoC] JtR SIMD support enhancements
Hi Lei,
On Sat, Mar 28, 2015 at 09:36:33AM +0800, Lei Zhang wrote:
> Currently there seems not much left to do for MIC
There is. We currently only have MIC intrinsics introduced for DES.
You should introduce them for whatever other formats currently have
intrinsics for SSE2 and such. magnum already told you where to start,
but he spoke in context of AVX2 and AVX-512. Well, for AVX-512 you'd
start with MIC intrinsics, which are mostly the same.
For AVX2, do you have where to test, or do you need access to our AVX2
machine? Actually, you can always test with Intel's SDE, but it may be
easier to test on the real thing - and you'd need that for benchmarks.
Another potential sub-task for MIC is to add OpenMP offload support.
Right now, we're only building for MIC fully natively. This runs fast,
but in typical installs it means that JtR's input and output files
reside in the MIC card's RAM, and will be lost if the card is rebooted.
(This may be corrected by mounting a host directory via NFS.) Also, to
use multiple MIC cards, multiple instances of JtR need to be managed
manually (or with scripts), whereas with OpenMP offload this might be
transparent to the user.
OTOH, I expect OpenMP offload to run significantly slower, except for
the slowest hash types like bcrypt where it'd run at roughly the same
speed as native. So ideally we'd support both fully native and offload
builds.
Thanks,
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.
