Vector2|3|4: Use Math.trunc() in roundToZero().

[ycw]

Math.trunc is much faster than current impl

[github-actions]

📦 Bundle size

Full ESM build, minified and gzipped.

Filesize dev	Filesize PR	Diff
646.4 kB (160.4 kB)	646.2 kB (160.3 kB)	-243 B

🌳 Bundle size after tree-shaking

Minimal build including a renderer, camera, empty scene, and dependencies.

Filesize dev	Filesize PR	Diff
439.2 kB (106.4 kB)	439 kB (106.4 kB)	-243 B

[WestLangley]

Math.trunc is much faster than current impl

FWIW, in my testing on Safari/FF MacOS -- it's marginally slower.

[ycw]

@WestLangley chrome116 on win11

[WestLangley]

Always test on a variety of OS/Browsers before making performance claims.

Just as in this case, results can be surprising. :-)

[Mugen87]

Tested on a macMini with M2 Pro:

Overall, Math.trunc() seems the more performant solution compared to what is implemented now. And it is more readable than the bitwise operation.

[Mugen87]

Let's give this a try!

[manthrax]

Wow.. blows my mind that Math.***( x ) is faster than an x | 0...