Intel MKL
From HP-SEE Wiki
(→Recommendations for Configuration and Usage) |
|||
Line 70: | Line 70: | ||
== Recommendations for Configuration and Usage == | == Recommendations for Configuration and Usage == | ||
- | |||
- | |||
- | |||
- | |||
- | |||
- | |||
- | |||
- | |||
- | |||
- | |||
- | |||
- | |||
- |
Latest revision as of 08:02, 5 September 2011
Contents |
- Web site: http://software.intel.com/en-us/articles/intel-mkl/
- Described version: 10.2
- Licensing: Non-commercial/Commercial
- User documentation: http://software.intel.com/en-us/articles/intel-math-kernel-library-documentation/
- Download: Intel non-commercial downloads
- Source code: Not available
Authors/Maintainers
Math Kernel Library is a provided by Intel Software Network. Different ways of support are available at Intel MKL support page
Summary
Intel Math Kernel Library (Intel MKL) is a library of optimized, extensively threaded math routines for solving large science, engineering, and financial computational problems. The library supports C/C++ and Fortran language APIs. The most of Intel MKL is threaded which enables more efficient usage of today’s multi-core processors and easier parallelization of developers application or implementation of hybrid programming model. Threading is performed through OpenMP, and the following mathematical domains are supported:
- Sparse Linear Algebra – Sparse BLAS, Sparse Format Converters, PARDISO Direct Sparse Solver, Iterative Sparse Solvers and Pre-conditioners
- Fast Fourier Transforms
- Optimized LINPACK benchmark
- Vector Math Library
- Statistics Functions – Vector Random Number Generators, Summary Statistics Library
- Cluster Support – ScaLAPACK, Cluster FFT
Intel MKL supports Linux, Windows and Mac OS operating systems. Installation on Linux platform is done through simple install script, which deploys provided packages on target machine. As it is developed by Intel processor manufactures, this library is highly optimized for usage on Intel processors, including processors with multiple cores. In case of other processors types it is usually better to use libraries provided by their manufactures (AMD Core Math Library (ACML) for AMD processors or IBM's Engineering and Scientific Subroutine Library (ESSL) for their line of CPUs).
In addition to commercial licensing model, Intel MKL is also available through the academic non-commercial license, which has the same features as the commercial product, but its use is limited to educational and academic purposes. Non-commercial license is an individual-use license, and each individual has to obtain it personally. Note, however, that this license cannot be used in the work for hire, i.e. university academic or research staff cannot use it for the research performed as a part of their regular activities for which they get salaries. Therefore, this license can be used to assess and evaluate the library itself, while the production use requires commercial license.
Features
http://software.intel.com/en-us/articles/intel-mkl/#details
Architectural/Functional Overview
- high level design info, how it works, performance - may be a link, or several links
Usage Overview
Intel® Math Kernel Library for Linux* OS User’s Guide
Using Code examples User's Guide page
Dependencies
n/a
HP-SEE Applications
- NUQG Numerical study of ultra-cold quantum gases
Resource Centers
Usage by Other Projects and Communities
- If any