Vectorized Execution

SUMMARY

Vectorized execution is a performance optimization technique that processes multiple data points simultaneously using CPU vector instructions, rather than processing one data point at a time. This approach significantly improves query performance by maximizing CPU efficiency and reducing overhead in database operations.

How vectorized execution works

Instead of processing data row by row, vectorized execution operates on blocks or vectors of data at once. This approach leverages modern CPU capabilities, particularly Single Instruction Multiple Data (SIMD) instructions, to perform operations on multiple values simultaneously.

This is particularly valuable for time-series databases where operations often need to be performed on large sequences of chronological data points.

Next generation time-series database

QuestDB is an open-source time-series database optimized for market and heavy industry data. Built from scratch in Java and C++, it offers high-throughput ingestion and fast SQL queries with time-series extensions.

Try live demo Read documentation

Benefits for time-series data processing

Vectorized execution offers several key advantages for time-series data:

Reduced CPU overhead: By processing multiple data points at once, there's less per-operation overhead
Better cache utilization: Working with blocks of data improves CPU cache efficiency
Improved throughput: Parallel processing of data vectors accelerates complex calculations

For example, when calculating moving averages or performing aggregations across time windows, vectorized execution can process entire blocks of timestamps and values simultaneously.

Next generation time-series database

Try live demo Read documentation

Implementation considerations

When implementing vectorized execution, several factors need to be considered:

Vector size optimization

The size of data vectors needs to be balanced carefully:

Too small: Doesn't fully utilize CPU capabilities
Too large: May cause cache misses and reduce efficiency

Data layout

Column-oriented storage often works better with vectorized execution as it allows for:

Contiguous memory access
Better compression
More efficient SIMD operations

Here's how vectorized execution might process a time-series aggregation:

This approach is particularly effective for operations like:

Calculating moving averages
Performing statistical analysis
Computing time-based aggregations
Applying mathematical transformations