Overview

PyArrow 17.0, released on August 21, 2024, brings performance improvements for Parquet reading and compute operations.

Main Features

Performance

Parquet reading is faster thanks to better parallelism and compute functions benefit from additional SIMD optimizations.

python
import pyarrow.parquet as pq
import pyarrow.compute as pc

# Optimized reading
table = pq.read_table('data.parquet')
print(f'Rows: {table.num_rows}')

# Fast vectorized compute
total = pc.sum(table.column('amount'))
print(f'Total: {total.as_py()}')

Sources