Path: news.mathworks.com!not-for-mail
From: <HIDDEN>
Newsgroups: comp.soft-sys.matlab
Subject: speed of matlab sparse against c++ sparse librariers
Date: Thu, 26 Feb 2009 22:00:04 +0000 (UTC)
Organization: The MathWorks, Inc.
Lines: 11
Message-ID: <go73d4$4i8$1@fred.mathworks.com>
Reply-To: <HIDDEN>
NNTP-Posting-Host: webapp-02-blr.mathworks.com
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 8bit
X-Trace: fred.mathworks.com 1235685604 4680 172.30.248.37 (26 Feb 2009 22:00:04 GMT)
X-Complaints-To: news@mathworks.com
NNTP-Posting-Date: Thu, 26 Feb 2009 22:00:04 +0000 (UTC)
X-Newsreader: MATLAB Central Newsreader 1337510
Xref: news.mathworks.com comp.soft-sys.matlab:521167


Hi,

I need to have a very solver for my FEM software.
For prototyping I've written all code in MATLAB and it's already quiet fast, however I want to have more performance.

Now I consider rewriting the solver in C++ with sparse matrix libraries as MTL.
Therefore I wondered what the performance gain of using a C++ sparse library over MATLAB would be. Will it be about the same, twice as fast etc.?

I'm just interested in performing some sparse matrix multiplications (nxn with n~30 000) and solve the system with preconditioned conjugate gradients. I'm not worried about the assembly of the FEM matrices.

Thanks a lot