Performance

By selecting the appropriate index type, typical search times can be kept within a few milliseconds. The performance of staticseek is influenced by various factors, including the size of the dataset, the complexity of the queries, and the chosen indexing strategy.

Search Performance for a 4MB Dataset

The following benchmarks illustrate the performance of different index types under worst-case scenarios, with approximately 100 articles:

Exact Match

LinearIndex: < 5ms
GPULinearIndex: < 5ms
HybridTrieBigramInvertedIndex: < 1ms

Fuzzy Search

LinearIndex: < 150ms
GPULinearIndex: < 25ms
HybridTrieBigramInvertedIndex: < 2ms

Index Generation

LinearIndex: ~1sec
GPULinearIndex: ~1sec
HybridTrieBigramInvertedIndex: ~2sec

Gzipped Index Size

LinearIndex: 1.3MB
GPULinearIndex: 1.3MB
HybridTrieBigramInvertedIndex: 0.5MB

Recommendations for Optimizing Performance

Choose the Right Index Type: For smaller datasets, LinearIndex may be sufficient. For larger datasets or applications requiring fuzzy search, consider GPULinearIndex or HybridTrieBigramInvertedIndex.
Pre-generate Indices: Generate indices during the build process to optimize search performance at runtime.

For detailed benchmarks across different hardware configurations and index types, see the Benchmarks section.