Smile v1.4.0 Release Notes
Release Date: 2017-08-06 // over 6 years ago-
๐ 1. Add smile-netlib module that leverages native BLAS/LAPACK for matrix computation. See below for the details how to enable it.
- Add t-SNE implementation. ๐ 3. Improve LLE and Laplacian Eigenmaps performance.
- Export DecisionTree and RegressionTree to Graphviz dot file for visualization.
- Smile shell is now based on Scala 2.12. ๐ 6. Bug fixes.
โก๏ธ To enable machine optimized matrix computation, the users should add
the dependency of smile-netlib:<dependency> <groupId>com.github.haifengl</groupId> <artifactId>smile-netlib</artifactId> <version>1.4.0</version> </dependency>
and also make their machine-optimised libblas3 (CBLAS) and liblapack3 (Fortran)
available as shared libraries at runtime.OS X
Apple OS X requires no further setup as it ships with the veclib framework.
๐ง Linux
Generically-tuned ATLAS and OpenBLAS are available with most distributions
๐ฆ and must be enabled explicitly using the package-manager. For example,- sudo apt-get install libatlas3-base libopenblas-base
- โก๏ธ sudo update-alternatives --config libblas.so
- โก๏ธ sudo update-alternatives --config libblas.so.3
- โก๏ธ sudo update-alternatives --config liblapack.so
- โก๏ธ sudo update-alternatives --config liblapack.so.3
๐ However, these are only generic pre-tuned builds. If you have an Intel MKL licence,
you could also create symbolic links from libblas.so.3 and liblapack.so.3 to libmkl_rt.so
or use Debian's alternatives system.๐ Windows
๐ The native_system builds expect to find libblas3.dll and liblapack3.dll
on the %PATH% (or current working directory). Besides vendor-supplied
implementations, OpenBLAS provide generically tuned binaries, and it
๐ is possible to build ATLAS.