[ Thread Index |
Date Index
| More lists.tuxfamily.org/eigen Archives
]
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] Eigen rocks
- From: Christian Mayer <mail@xxxxxxxxxxxxxxxxx>
- Date: Fri, 27 Feb 2009 18:38:36 +0100
- 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=kAx8gkZdkX0E/foSa204BR6O/Ht2XzsAO8BBgLtLg5A=; b=R4gPcQixN1nEHIZg62zmzt9xK4b9JpZK22YiNVil6FUt1o2OxWhbzTMb2VYM1L64q6 VJAZPsJx6sXd/AmUK9WXWn5LNzVr7USYQ6jEZipZMfPQYB4XQSE5z52MOt8iYqvf0Zs9 4aB+Rq4PD0ssCxswvMOSEqiqwjuxYZKXLlWoA=
- 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=JrU0hVzS9V62PeSglXdk1MczL0xLnEznG0SXpqRUQTxh+vwKKNCqd4sNxhKHvd/3Wv DMHtLBJz++exMGCmA7ug5lQOE6saqDuO0e6u3QZyc7ag93pJPaw6ntxWPKlWgcBEJlIi K5fet8VtVTqUpi0Rm/8u3X0ELDaN3fpT9ThQ0=
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256
Rohit Garg schrieb:
>> For 1), Eigen itself doesn't do any multithreading, but on the plus
>> side, Eigen is stateless so completely (trivially) thread-safe.
> Finally, I come across something that looks like it built for the 21st
> century from day 1. I love you lot. Look here
>
> http://rpg-314.blogspot.com/2009/01/thread-safety.html
Getting a bit OT:
Although it's a shame that OpenGL isn't thread save, it doesn't matter.
It's supposed usage in a multithreaded environment is that you've got
one thread that talks to OpenGL and that is getting it's commands from
the other threads. And as most programs don't care about raw OpenGL but
prefer a scenegraph (any program that's not having trivial complexity
usually needs it) you can easily use a multithreaded scenegraph and your
requirements are satisfied.
Multithreading in Eigen is a diffentent thing. Using only small matrices
you don't want a math library to use multithreading - that overhead
would be too intense. If it comes to *big* matrices (big enough so that
they aren't fitting in the cache) then it's great to have multithreading
used by a math lib.
In both cases the math lib should be thread safe, so that *you* can use
it for multithreading.
Eigen is thread safe, so the basic requirement is fullfilled :)
As Eigen is optimized to handle small, fixed size matrices in the best
possible way, any multithreading support would be slowing it down.
And for big matrices I'd use MKL
..
CU,
Christian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
iEYEAREIAAYFAkmoJRwACgkQoWM1JLkHou3HfACcDo01CSmo/DAsa3x8E/NW8XVX
DBYAoIVtbyYoGqkM9y5EJ20nWsWfceNT
=F2LJ
-----END PGP SIGNATURE-----