|Re: [eigen] FLENS C++ expression template Library has excellent documentation|
[ Thread Index |
| More lists.tuxfamily.org/eigen Archives
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] FLENS C++ expression template Library has excellent documentation
- From: Christian Mayer <mail@xxxxxxxxxxxxxxxxx>
- Date: Mon, 20 Apr 2009 16:59:47 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:sender:message-id:date:from :reply-to:user-agent:mime-version:to:subject:references:in-reply-to :x-enigmail-version:content-type:content-transfer-encoding; bh=MX3ouOFVBlcBL1imtorWXtQkuwKAVXVBe3hlqDNHTMg=; b=rero1F3nlpxjpo9ZIapqzysZxD+LkT3/7pZScWeIIv2LVm8p/e2e3JTIBAoI3L7vyy UXfMGvNoJuh0rbC1fTOnWJSwvbzbn+Zq9pbNWA3uKTIOP8d9GAPDHjiOfQzNL0d5+O91 ri/Xy4+Oq4hUmN92D9aZnBBdRkB+KvQF6E+Dk=
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=sender:message-id:date:from:reply-to:user-agent:mime-version:to :subject:references:in-reply-to:x-enigmail-version:content-type :content-transfer-encoding; b=AuAnZpGVk3xRF+6MfGzQXAFw3PNOUGugC0SUEcB5UlMgysS5Yn987RRgfj3z3+vRWX gUMzTd5gUHPuMCP1k/boVjiwJVABF/lMVohz4obiWKyBoUdPOe/qkDy//LhM/LkvXKqa +ia4v0KAkC2sKtgLKh1hWZ51zur9rkIX1wWqU=
-----BEGIN PGP SIGNED MESSAGE-----
Helmut Jarausch schrieb:
> On 18 Apr, Christian Mayer wrote:
>> (disclaimer: I haven't done much parallelization in the past, but I
>> listened with much interest a few lectures and different discussions on
>> the net)
>> Do NOT use OpenMP in our case!
>> OpenMP is great to parallelize a few loops in old code where you can't
>> spend the time to do it right. If you've got the choice you should
>> allways rethink every algorithm and implement it in a parallel way (with
>> the threading lib of your choice)
> Have you had a look at OpenMP 3 (offered by gcc-4.4, e.g.).
No, I didn't. Development might have gone on and also my experience with
multithreading might not be big enough to make the best decision (see
the disclaimer :)
> It has the "task" directive which offers some freedom especially
> for C++ like programm. OpenMP has the advantage of being an accepted
Perhaps it might be the solution (I still didn't have a look at it...).
But you should allways consider, that a generic task handling and
scheduler might not be able to give you the power (speed wise and
complexity wise) that an optimal coded sheduler might offer.
In the end it boils down to the so typical:
vs. compiling speed
vs. runtime speed
vs. code complexity
vs. cross platform capabilities.
There eigen has gone it's very special path that I think is great for
this "small" libaray. Choosing the right multithreading environment
should consider to follow and extend the current path and not be
distraced by the ease of implementation.
Anyway, it's an open source project, so the person implementing the
stuff is the person to decide :)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
-----END PGP SIGNATURE-----