Skip to content
Tony Wilkes edited this page Apr 19, 2025 · 50 revisions

User contributed benchmarks involving fastverse packages. Please add yours!

Desirable benchmark features

ID Feature Name
S Different Data Sizes (Number of Rows)
T Different Data Types (Integer, Double, String, etc.)
C Different Cardinality (Number of Groups)
R Real Data (Not Generated)
L Different Languages (R, Python, etc.)

General Benchmarks

Title + Link System Date Author Features
Database-like ops benchmark Linux Server 256Gb (E6i) Recurrent DuckDB Labs S T C L
DataFrame package benchmark Windows Server 256Gb June 2024 Adrian Antico S T C R L
collapse, arrow, and data.table (source) M1 MacBook Pro 16Gb August 2022 Sebastian Krantz T C R
collapse and dplyr Windows Laptop 8Gb January 2021 Sebastian Krantz S T R
collapse intro benchmark Windows Laptop 8Gb August 2020 Sebastian Krantz T R

Specific Functionality Benchmarks

Title + Link System Date Author Features
Benchmarking rolling window functions in R - October 2024 Mikkel Roald-Arbøl C
Joins: collapse vs. data.table - 2024 Zander Prinsloo T C
300k p-value simulation - November 2023 Andrew Ghazi
Panel-lags and demeaning (slide 40) M1 MacBook Pro 16Gb June 2022 Sebastian Krantz R
Unique values: collapse vs. data.table vs. dplyr - December 2020 John MacKintosh T R
Vector processing benchmarks - May 2023 Krzysztof Dyba L
Raster processing benchmarks - May 2023 Krzysztof Dyba L
broadcasted addition ('+') of decimal numeric arrays: 'broadcast' ('R') vs 'Numpy' ('Python') Windows 11, 32GB RAM, 2.50 GHz April 2025 Tony Wilkes S C L