The Lab for Scientific Computing (LSC) at the University of Notre Dame
announces the initial public release of the Matrix Template
Library (MTL), a library of generic components for high performance
numerical linear algebra.
The Matrix Template Library is:
Generic:
MTL uses the generic programming paradigm to allow algorithms to be
expressed in a data-format-independent manner. Generic programming
techniques are also used to characterize and express MTL's
performance optimizations.
Orthogonal and Comprehensive:
MTL includes a large number of data formats and algorithms,
including most popular sparse and dense matrix formats and
functionality equivalent to BLAS level 3.
Flexible and Extensible:
The MTL generic programming interface is designed to allow new
functionality to be easily incorporated.
High Performance:
MTL provides performance equal to (and in many cases better than)
vendor-tuned math libraries.
Portable:
MTL is written entirely in ANSI C++ and can be compiled and used on
any target platform with an ANSI C++ compiler. No external
libraries are required.
Reliable:
MTL is extensively tested on a nightly basis. Over 350,000
separate tests are conducted for each supported configuration.
A scoreboard of test results is mainrained on the MTL web site.
MTL source code and documentation is available from
http://www.lsc.nd.edu/research/mtl/
This archive was generated by hypermail 2b29 : Wed Feb 20 2002 - 03:20:08 EST