Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 29 Jan 2014 15:42:48 +0000
From:      David Chisnall <theraven@FreeBSD.org>
To:        Michael Schmiedgen <schmiedgen@gmx.net>
Cc:        Dmitry Marakasov <amdmi3@amdmi3.ru>, FreeBSD Current <freebsd-current@FreeBSD.org>
Subject:   Re: [OT] ta-spring
Message-ID:  <57D26185-3443-4EA4-822E-D6D50D23551F@FreeBSD.org>
In-Reply-To: <52E92024.9010501@gmx.net>
References:  <20140129022743.GO52820@hades.panopticon> <7B2C5E6B-C845-4AFB-845F-AAF6F3C1F239@FreeBSD.org> <20140129144518.GR52820@hades.panopticon> <52E91955.9030406@gmx.net> <B93982A8-3151-4A30-89C9-74F83BEF9D71@FreeBSD.org> <52E92024.9010501@gmx.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On 29 Jan 2014, at 15:37, Michael Schmiedgen <schmiedgen@gmx.net> wrote:

> On 29.01.2014 16:16, David Chisnall wrote:
>> On 29 Jan 2014, at 15:08, Michael Schmiedgen <schmiedgen@gmx.net> =
wrote:
>>=20
>>> Can we expect a current version of spring in ports soon? That would
>>> be nice! AFAIK newer versions require OpenMP. Will this compile with
>>> our (new 3.4 soon) base clang?
>>=20
>> Base clang doesn't support OpenMP.  We should probably import Intel's =
Clang fork into ports:
>>=20
>> http://clang-omp.github.io
>>=20
>> This can then be used to compile things that need both libc++ and =
OpenMP.  Intel's OpenMP runtime is permissively licensed now, but will =
likely require a small amount of porting to get it to work on FreeBSD =
(it supports Linux and OS X).
>>=20
>=20
> I thought OpenMP will be an integral part of LLVM/clang in near =
future,
> at least the front-end part? It seems there are plans to even =
integrate
> the runtime in the llvm project source tree:
>=20
> http://openmp.llvm.org/
>=20
> Ok, so llvm/clang 3.4 obviously will not ship with OpenMP, but maybe
> later versions.

Active development happens in Intel's tree, and is slowly being merged =
upstream.  Eventually, Clang will have full OpenMP 4 support, but =
Intel's tree will have it first and there is likely to be a lag before =
it makes it into mainline clang.

As such, it would make sense to have a port as a stop-gap until it is =
ready.

David




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?57D26185-3443-4EA4-822E-D6D50D23551F>