Armadillo + Blas LAPACK/+ openblas

Armadillo is a mathematical computing library that can be computed using different mathematical libraries.

First, download armadillo,4.400.pdf at http://arma.sourceforge.net/download.htmlto use vs2012for compilation.

After the download, go to the producer & LAPACK (below) Library.

Then, follow the instructions in readme.txt to delete the two annotations in config. HPP to enable acceleration for other computing libraries ..

According to tests on the internet, openblas is highly efficient and mil is also very good, but the latter is a commercial library and is charged.

 

Use 64-bit Blas & LAPACK 3.2

1. Select the compilation platform, x64

2. Add the include directory,/include

3. Add the reference directory, that is, the folder with blas_win64_mt.lib, examples/lib_win64

4. Add the Lib in the 3 folder to project options-connector-input-Additional dependency.

5. Place the DLL in the 3 folder to the project directory.

5. Compile and run.

How to Use openblas

The new binary package version only contains the. A file, but not the. Lib file. The old version has a. Lib file.

Openblas reference

The author's article, [important] The author detailed each step of operation: https://github.com/xianyi/OpenBLAS/wiki/How-to-use-OpenBLAS-in-Microsoft-Visual-Studio

This is the author's Q & A http://my.oschina.net/u/274748/blog/101897

And this Q & A http://my.oschina.net/u/274748/blog/158728

 

Use liband dllof 0.2.820.x86. the name is openblas-v0.2.8-x86-win.zip.

Efficiency Comparison:

Calculate the time consumption of 1000 × 1000 matrix inverse, release version

64-bit Blas & LAPACK	Openblas
8200 + MS	250 ms

Efficiency is fast.

Armadillo + Blas LAPACK/+ openblas