terser-performance

Repo to benchmark performance of Terser.

Usage

git clone https://github.com/filipesilva/terser-performance.git
npm install
npm test

You should see something similar to this:

$ npm test

> terser-performance@1.0.0 test D:\sandbox\terser-performance
> npm run benchmark && npm run profile


> terser-performance@1.0.0 benchmark D:\sandbox\terser-performance
> npm run benchmark-medium && npm run benchmark-large


> terser-performance@1.0.0 benchmark-medium D:\sandbox\terser-performance
> benchmark -- node run-terser.js ./medium-sample.js

[benchmark] Benchmarking process over 5 iterations, with up to 5 retries.
[benchmark]   node run-terser.js ./medium-sample.js (at D:\sandbox\terser-performance)
[benchmark] Process Stats
[benchmark]   Elapsed Time: 3953.00 ms (3745.00, 3768.00, 4128.00, 4152.00, 3972.00)
[benchmark]   Average Process usage: 1.00 process(es) (1.00, 1.00, 1.00, 1.00, 1.00)
[benchmark]   Peak Process usage: 1.00 process(es) (1.00, 1.00, 1.00, 1.00, 1.00)
[benchmark]   Average CPU usage: 10.09 % (8.38, 9.37, 12.71, 11.47, 8.54)
[benchmark]   Peak CPU usage: 51.56 % (45.40, 48.40, 65.50, 53.20, 45.30)
[benchmark]   Average Memory usage: 124.49 MB (123.21, 121.55, 130.48, 123.46, 123.75)
[benchmark]   Peak Memory usage: 158.65 MB (157.22, 155.60, 160.69, 160.14, 159.61)

> terser-performance@1.0.0 benchmark-large D:\sandbox\terser-performance
> benchmark -- node run-terser.js ./large-sample.js

[benchmark] Benchmarking process over 5 iterations, with up to 5 retries.
[benchmark]   node run-terser.js ./large-sample.js (at D:\sandbox\terser-performance)
[benchmark] Process Stats
[benchmark]   Elapsed Time: 29140.20 ms (25791.00, 28231.00, 30336.00, 30615.00, 30728.00)
[benchmark]   Average Process usage: 1.00 process(es) (1.00, 1.00, 1.00, 1.00, 1.00)
[benchmark]   Peak Process usage: 1.00 process(es) (1.00, 1.00, 1.00, 1.00, 1.00)
[benchmark]   Average CPU usage: 13.52 % (11.48, 12.30, 15.04, 13.91, 14.88)
[benchmark]   Peak CPU usage: 98.74 % (112.50, 76.50, 112.50, 68.80, 123.40)
[benchmark]   Average Memory usage: 425.00 MB (422.54, 426.24, 422.82, 424.08, 429.33)
[benchmark]   Peak Memory usage: 589.08 MB (588.97, 588.98, 588.83, 592.81, 585.82)

> terser-performance@1.0.0 profile D:\sandbox\terser-performance
> npm run profile-medium && npm run profile-large


> terser-performance@1.0.0 profile-medium D:\sandbox\terser-performance
> node run-terser.js medium-sample.js CPU_PROFILE && node run-terser.js medium-sample.js HEAP_PROFILE


Profiler: taking CPU profile

Profiler: saved CPU profile to D:\sandbox\terser-performance\PROFILE_20191001T113704685Z.cpuprofile

Profiler: taking heap profile

Profiler: saved heap profile to D:\sandbox\terser-performance\PROFILE_20191001T113709715Z.heapprofile

> terser-performance@1.0.0 profile-large D:\sandbox\terser-performance
> node run-terser.js large-sample.js CPU_PROFILE && node run-terser.js large-sample.js HEAP_PROFILE


Profiler: taking CPU profile

Profiler: saved CPU profile to D:\sandbox\terser-performance\PROFILE_20191001T113737368Z.cpuprofile

Profiler: taking heap profile

Profiler: saved heap profile to D:\sandbox\terser-performance\PROFILE_20191001T113818837Z.heapprofile

The output prefixed with [benchmark] shows time, cpu, and memory resource usage.

The .cpuprofile and .heapprofile files contain a detailed profiled that can be opened in Google Chrome. To see the profiles, open chrome://inspect/#devices on Chrome, then click Open dedicated DevTools for Node. The .cpuprofile file shows how time was used per function and should be loaded in the Profiler tab. The .heapprofile file shows how memory was used per function and should be opened in the Memory tab. After loading them you can zoom, pan, and mouseover individual functions for more details. There is also different kinds of views available on the downwards arrow by Chart on the topleft.

You can use git clean -fx to clean up all the ignored output files after a benchmark. The node_modules folder is ignored but won't be deleted.

Individual benchmark tasks

This repository contains two sample files to use with Terser: medium-sample.js and large-sample.js.

The following npm scripts benchmark Terser performance with those two files.

• npm run benchmark-medium
• npm run benchmark-large
• npm run benchmark (runs both above)
• npm run profile-medium
• npm run profile-large
• npm run profile (runs both above)