Starting with LAPACK 3.3.0, the LAPACK team decided to join forces with Kitware Inc. To develop a new build system for LAPACK using CMake 2. This fruitful collaboration gave LAPACK the ability to easily compile and run on the popular Linux, Windows, and Mac OS/X platforms in.
The recipe is valid with CMake version 3.5 (and higher) and has been tested on GNU/Linux, macOS, and Windows. Many numerical codes rely heavily on matrix and vector operations. Think for example of matrix-vector and matrix-matrix products, the solution of linear system of equations, the calculation of eigenvalues and eigenvectors or singular. This makes shared libraries, but I'm now struggling to get CMake to link my local shared library to these (it keeps wanting to look at the archive libraries instead). I might now try instead pulling all the.o s from blas and lapack into my local shared library. That's ugly but hopefully should work. Suggestions welcome. LAPACK has been thoroughly tested, on many different types of computers. The LAPACK project supports the package in the sense that reports of errors or poor performance will gain immediate attention from the developers.
The LAPACK community has decided to extend its support to Microsoft Windows based users. The decision was taken due to the large amount of requests we received.The project included in this initiative are LAPACK, CLAPACK and ScaLAPACK . For more information on those libraries, please refer to their website.Computational Software Development with Windows HPC
TheInnovative Computing Laboratory (ICL) at the University of Tennesseehas been engaged in High-performance computing (HPC) research throughthe development of applications, scientific codes, and computationallibraries that have helped form the fabric of HPC computing in the 21stcentury.
One of our goal is to extend HPC applications, software andmathematical libraries to the Microsoft Windows Compute Clusterenvironment. To this end we hope to understand the requirements andopportunities for computational based research and innovation utilizingWindows high performance computing.
What you will find here
Here you should be able to able all Windows-related libraries and package for LAPACK (Fortran), CLAPACK (C), and ScaLAPACK (C and Fortran).The first question to ask you is what you really need .
- For a basic usage, we recommend to use prebuilt libraries.
- For an advanced usage requiring special compilation or optimization, we recommend to use the packages that will allow you to compile our libraries on your machine.
Pre-built libraries
This is the easy way to get the library installed on your machine.Read carefully the requirements that are needed.Cmake Lapack Mkl
Install packages
This is the advanced way to get the library installed on your machine. It will require you to compile the library.Read carefully the requirements that are needed.Step by step to run example
For each library, we are providing a quick step-by-step tutorial to get you started. Our aim is to run LAPACK natively on Windows, that is not using cygwin but Microsoft Visual Studio.Those step by steps have been mostly designed and/or imporved by our users, so feel free to send a feedback or contribution.Cmake Lapack Project
Other ICL projects ported to Windows
PLASMA (The Parallel Linear Algebra for Scalable Multi-core Architectures = LAPACK + Multi-core) project has released a binary installer for 32 bit and 64 bit PLASMA libraries. It includes statically compiled PLASMA libaries, and testing, timing and example executables. Use of the PLASMA libraries requires a C compiler (e.g. MS VC) and a BLAS library (e.g. Intel MKL, AMD ACML).Acknowledgments
Thank you to the LAPACK community, especially the LAPACK forum community. Your requests, comments, contribution are greatly appreciated and we are doing our best include them in our libraries.
Thank you to Microsoft Corporation for helping us and providing us expertise on Windows platform.
Thank you to Kitware for helping designing the CMAKE packages.
At some point, finding BLAS and Lapack for Intel compilers on Windows seems to have broken. This issue occurs at least for CMake 3.16.8 and CMake 3.18-rc3. This happens for Intel Parallel Studio XE as well as the new Intel oneAPI compilers.
It looks at first glance like there may be an issues with properly escaping the compiler path that contains spaces–i.e. this may not be exactly a 'FindBLAS' issue, but maybe an issue in general for this compiler.
CMakeLists.txt
Results for:
Cmake Lapack Link
I am happy to provide additional info etc.
