Skip to content

BENCHMARK.md

Latest commit

 

History

History
65 lines (47 loc) · 3.01 KB

BENCHMARK.md

File metadata and controls

65 lines (47 loc) · 3.01 KB

Event Loop Starvation Benchmark

This benchmark measures event loop blocking/starvation when processing JSON at different data volumes, validating the statistics mentioned in the main README.md.

Overview

The benchmark tests both built-in JSON methods and yieldable-json async methods to measure how long each blocks the event loop during parse and stringify operations. It uses nanosecond precision measurements to accurately detect sub-millisecond blocking behavior.

Test Data

Test data is generated dynamically using Faker.js before each benchmark run. This eliminates the need for static JSON fixture files and provides more realistic test data.

Data Sizes

Data is generated at the following target sizes:

  • 230 KB (small)
  • 654 KB (medium)
  • 2.6 MB (large)
  • 4.4 MB (xlarge)

Payload Types

Each size is tested with five different payload types:

  • nested-objects: Deeply nested object structures with realistic user/profile data
  • array-heavy: Large arrays of product/commerce data
  • string-heavy: Objects with long text content (articles, descriptions)
  • mixed: Combination of users, products, orders, companies, and posts
  • toJSON-heavy: Objects with custom toJSON() methods

Running the Benchmark

npm run benchmark

Output

The benchmark produces detailed results for each payload type, plus a summary table showing worst-case delays:

Summary Results (Worst Case Across All Payload Types)

Data Volume Operation JSON [ms] Yieldable [ms]
230 KB parse 1.54 2.87
230 KB stringify 0.66 6.60
654 KB parse 3.94 16.24
654 KB stringify 1.95 1.48
2.6 MB parse 13.35 35.72
2.6 MB stringify 9.86 2.59
4.4 MB parse 19.81 38.32
4.4 MB stringify 18.85 4.75

Note: Times are maximum event loop delays (worst case) measured in nanoseconds and displayed in [ms] with 2 decimal precision. Benchmarked February 2026.

Key Insights

✅ yieldable-json excels at stringify operations for large data:

  • At 2.6 MB: Native JSON blocks for 9.86ms, yieldable-json only 2.59ms (73% faster)
  • At 4.4 MB: Native JSON blocks for 18.85ms, yieldable-json only 4.75ms (75% faster)

✅ Consistent low blocking times: yieldable-json maintains predictable ~2-5ms blocking for stringify regardless of data size

⚠️ Parse operations on large data:

  • Native JSON is still faster for parse operations on very large data due to yielding overhead
  • yieldable-json parse prevents complete event loop starvation and maintains responsiveness
  • For parse-heavy workloads with massive data, consider native JSON if absolute speed matters more than concurrency
  • yieldable-json is ideal for stringify-heavy workloads and high-concurrency scenarios