Getting Started

Install and benchmark learned indexes in your PostgreSQL database

Installation

# Clone and build

git clone https://github.com/omendb/pg-learned

cd pg-learned

cargo pgrx install

# In PostgreSQL

CREATE EXTENSION omendb;

Compare learned indexes vs B-trees

Get extension version info

Learn how learned indexes work

Typical results when testing the PostgreSQL extension on sequential data:

Example Output from learned_index_benchmark(50000):

Dataset: 50,000 sequential keys

Function: learned_index_benchmark(50000)

Performance Comparison:

Linear Index: 6,500,000 queries/sec (50,000 found)

RMI Index: 7,200,000 queries/sec (50,000 found)

BTreeMap: 2,700,000 queries/sec (50,000 found)

Linear Speedup: 2.4x

RMI Speedup: 2.7x

Results vary based on data distribution and hardware.

Sequential/time-series data shows best performance gains.

Test yourself: Install the extension and run SELECT learned_index_benchmark(10000) to see performance on your hardware and data.

-- Get version info

SELECT learned_index_version();

-- Learn about the technology

SELECT learned_index_info();

-- Benchmark with 50,000 keys

SELECT learned_index_benchmark(50000);

This extension is just the beginning. We're building a full database from the ground up.