Overview of MTL4
Today everything can be done nicely on a computer except one thing:
Computing!
-- Our experience
- Not worrying about technical details speeds up scientific and engineering software development tremendously.
- Not hard-wiring hardware features in application allows for easy porting to new platforms.
High
PerformanceHigh
Productivity
HPC programming made easy with MTL4!
Features
Open-Source Edition |
Supercomputing Edition |
|
|
|
The Matrix Template Library 4 |
The Matrix Template Library is:
-
Generic, Orthogonal, and Comprehensive: MTL4 includes a large number of data formats and algorithms, including most popular sparse and dense matrix formats and functionality.
-
Flexible and Extensible: The MTL4 interface is designed to allow new functionality to be easily incorporated.
-
High Performance: Benchmarks with MTL4 on current compilers showed very good performance on the sequential Open-Source Edition as well as on the parallel Supercomputing Edition.
-
Reliable: MTL 4 is tested nightly on several compilers.
- Portable: MTL 4 is written in ANSI C++ and can be compiled and used on any target platform with an ANSI C++ compiler. Currently, compilers tested every night are:
- Linux & MacOS
- clang++ 3.0 (32 bit)
- g++ 4.1.3 (32 and 64 bit)
- g++ 4.2.4 (64bit)
- g++ 4.3.4 (32 and 64 bit)
- g++ 4.4.3 (32 and 64 bit)
- g++ 4.5.1 (64 bit)
- g++ 4.6.0 (64 bit)
- g++ 4.6.1 (64 bit)
- icc 9.1 (32 and 64 bit)
- icc 10.0 (32 and 64 bit)
- icc 10.1 (32 and 64 bit)
- icc 11.0 (32 and 64 bit)
- icc 11.1 (32 and 64 bit)
- Windows
- VC 8.0 from Visual Studio 2005 (32 bit)
- VC 9.0 from Visual Studio 2008 (32 bit)
- VC 10.0 from Visual Studio 2010 (32 bit)
More editions under development:
- Multigrid Edition
- GPU Edition
- Accelerated Dense Edition
- BLAS interfaces
- LAPACK interfaces