nodejs-package-benchmark

This package allows you to benchmark different runtimes using popular packages operations.

Supported packages

• fastify - Temporary disabled.
• lodash
• prettier
• babel
• moment
• dotenv
• pinojs
• winston
• underscore
• typescript

Install

1npm i -g nodejs-package-benchmark

Comparison

To compare binaries, you can use the bench-it script.

Syntax: bench-it $BINARY ["baseline"]

This script allows you to compare the performance of binaries. If it's your first run, you need to generate the baseline data using:

1$ bench-it ./node baseline

To compare subsequent runs, simply omit the "baseline" option:

1$ bench-it ./node

[!TIP] It's recommended to have colordiff installed for a clearer comparison of differences.

Single run

To a pretty terminal output, run index.js

1$ node index.js
2cpu: 13th Gen Intel(R) Core(TM) i5-13600K (20 cores)
3node: v20.13.1 (/home/hzk/.nvm/versions/node/v20.13.1/bin/node)
4os: Linux 5.15.133.1-microsoft-standard-WSL2 x64
5---------------------------------------------------------------------
6babel
7  transform (code=true ast=true):                                    147.1 (15 samples)
8  transform (code=false):                                            150.4 (16 samples)
9dotenv
10  config:                                                            64.39K (6440 samples)
11lodash
12  .chunk:                                                            19.62M (1961973 samples)
13  .groupBy:                                                          6.303M (630542 samples)
14  .includes:                                                         18.38M (1837801 samples)
15  .orderBy:                                                          1.737M (173672 samples)
16moment
17  format (full):                                                     878.8K (87908 samples)
18  format:                                                            839.3K (83928 samples)
19  fromNow (YYYYMMDD):                                                153.4K (15339 samples)
20  subtract (10):                                                     288.7K (28871 samples)
21pinojs
22  info (10x):                                                        243.7K (24686 samples)
23prettier
24  format:                                                            71.71 (10 samples)
25  format (singleQuote=true useTabs=true):                            68.52 (10 samples)
26  format (semi=false):                                               70.61 (10 samples)
27typescript
28  transpile:                                                         85.23 (10 samples)
29  createSourceFile:                                                  1.866K (187 samples)
30  getSemanticDiagnostics:                                            2.102 (10 samples)
31underscore
32  .chunk:                                                            4.606M (467160 samples)
33  .groupBy:                                                          1.66M (169265 samples)
34  .includes:                                                         7.626M (762632 samples)
35  .orderBy:                                                          733.5K (73352 samples)
36winston
37  info (10x):                                                        43.13K (4431 samples)
38---------------------------------------------------------------------

To store it as JSON, just pipe output to a file:

1$ node index.js > results.json
2$ cat result.json
3[
4  {
5    "name": "babel",
6    "method": "benchmarkjs",
7    "operations": [
8      {
9        "name": "transform (code=true ast=true)",
10        "opsSec": 67.80076532539411,
11        "samples": 3
12      },
13      {
14...