[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20141218210940.GA7060@openwall.com>
Date: Fri, 19 Dec 2014 00:09:40 +0300
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: ARM NEON
On Thu, Dec 18, 2014 at 09:43:35PM +0100, magnum wrote:
> On 2014-12-18 20:55, magnum wrote:
> > On 2014-12-18 20:42, Solar Designer wrote:
> >> I think the current configure script won't detect NEON on ARM, even
> >> though it does detect SIMD extensions on x86. I think it should also
> >> detect NEON and pass -mfpu=neon then (like linux-arm32le-neon in
> >> Makefile.legacy does). Is this something you can add?
> >
> > https://github.com/magnumripper/JohnTheRipper/issues/911
>
> I take it this isn't added by -march=native?
There's no -march=native on ARM, at least not with the version of gcc I
have here.
> Then we need intrinsics to test. I guess we can find them in DES_bs.b.c.
Yes, or we may detect NEON in some other way, e.g.:
https://mailman.videolan.org/pipermail/vlc-commits/2011-July/007663.html
> I presume -mfpu=neon will result in the compiler defining the
> __ARM_NEON__ macro.
Right.
solar@...aro-ubuntu-desktop:~$ gcc -march=native -dM -E - < /dev/null | fgrep NEON
cc1: error: bad value (native) for -march switch
solar@...aro-ubuntu-desktop:~$ gcc -mfpu=neon -dM -E - < /dev/null | fgrep NEON
#define __ARM_NEON__ 1
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.
