NEWS.md

In data.table.threads: Analyze Multi-Threading Performance for 'data.table' Functions

data.table.threads v1.0.1

New Features

• Added a function that computes and adds recommended efficiency plot data (speedup lines and thread count points) based on the recommendedEfficiency argument (has been removed from findOptimalThreadCount). This allows the user to construct plots without recommended efficiency lines and points, and on the other hand, to easily add them when needed with the help of this function.
• findOptimalThreadCount now has recommendedEfficiency, an argument for adjusting the slope of the recommended speedup trend/line and the recommended thread count designating point in speedup plots generated by using the plot method on an object returned by the function.
• The data used for plotting (consisting of speedup trends/lines and key points) is now accessible via attributes of the data.table returned by findOptimalThreadCount.
• Added new arguments verbose and times to runBenchmarks and findOptimalThreadCount, allowing users to enable/disable detailed message printing and specify the number of benchmark repetitions.
• S3 plot method constructs a more intuitive ggplot for data_table_threads_benchmark class objects.
• Only user-facing functions are exported.
• Examples have been updated to use namespace qualification.
• Added a function to set the thread count for data.table operations based on recommended/optimal performance (in benchmarks) of a user-specified data.table function.

Improvements

• Extensively refactored my codebase to remove redundancy.
• Made several optimizations (using data.table over data.frame, implementation of faster data aggregation operations, minimizing overhead for various function calls, etc.).

data.table.threads documentation built on April 3, 2025, 10:08 p.m.