Oracle Machine Learning for R (OML4R) now supports Oracle R Distribution (ORD) version 4.4.1, which brings significant new features and improvements since version 4.0.5. ORD is Oracle’s redistribution of open source R that further improves R’s performance by dynamically linking to optimized, multi-threaded BLAS libraries, using all available cores and processors unlike standard R. Built upon the foundation provided by the R Core Team and the broader R development community, open source R 4.4.1 focuses on enhancing developer productivity and system performance with several key updates.

The updates in open source R 4.4.1 deliver significant performance improvements, with optimizations for graphics rendering, memory management, and missing data handling providing faster execution and reduced memory overhead. R 4.4.1 also introduces a native pipe operator that streamlines data transformations and makes code more readable and efficient, along with enhanced developer experience through simplified user-defined function (UDF) creation, improved data sorting and string handling, and an upgrade to C++17 as the default standard.

Compatibility

Oracle R Distribution 4.4.1 maintains compatibility with open source R 4.4.1 and is certified with Oracle Machine Learning for R (OML4R) 2.0. To use OML4R with ORD 4.4.1, customers should uninstall their current ORD and OML4R installations, then install ORD 4.4.1 followed by OML4R 2.0. Users upgrading from previous versions of R should recompile their third-party or custom packages under ORD 4.4.1.

Performance

The recent version of open source R, R 4.4.1, has delivered significant performance improvements. Compared to R 3.6.1, R 4.4.1 dramatically boosts base R performance, achieving up to 2.5 times faster results in our benchmark testing. Oracle R Distribution 4.4.1 builds on these improvements and delivers additional performance advantages through Intel MKL hardware-optimized math libraries. MKL provides substantial performance gains for vector and matrix operations, scaling effectively with additional threads across multi-core systems. Compared to open-source R, computational workloads including matrix operations, linear algebra, mathematical functions, and statistical analysis see significant speedups, with matrix multiplication and principal components analysis showing particularly impressive results. This release also includes performance improvements from enhanced graphics rendering, improved memory management, and optimizations to core R functions.

Benchmark results in Figure 1 demonstrate the performance improvements. For example, matrix multiplication runs over 50x faster when comparing MKL optimization and R’s internal BLAS/LAPACK libraries; with Linear Discriminant Analysis showing speedups of over 25x faster. These results demonstrate speedups across most computational workloads. In the chart, the blue bars represent ORD with R’s internal libraries (Netlib), while other bars show ORD with Intel MKL optimization across different thread counts.
 

Oracle R Distribution 4.4.1 benchmarks

Resources

Availability

Oracle R Distribution 4.4.1 is available on Oracle Linux yum addons channel. Complete details on all R changes and enhancements can be found in the official R-NEWS for versions 4.0.5 through 4.4.1.