Rounding & Precision Policy

Internal precision

All calculations use IEEE 754 double-precision floating point, providing roughly 15–17 significant decimal digits. Conversion factors are stored at full standard precision (e.g. 1 in = 0.0254 m exactly).

Display precision

• Default: up to 6 significant figures, trailing zeros suppressed.
• Very small or very large magnitudes automatically switch to scientific notation.
• Currency is displayed at 2 decimal places (4 for highly granular pairs).
• Counting units (bytes, items) are displayed as integers when the value is exact.

Rounding rule

We use round-half-to-even (banker's rounding), which is the IEEE 754 default and reduces statistical bias compared with always rounding half up.

Significant figures vs decimal places

• Significant figures describe the meaningful digits in a number regardless of scale — appropriate for scientific and engineering work.
• Decimal places describe digits after the decimal point — appropriate for financial and tabular work.

Where a unit category has a clear convention (e.g. currency uses decimal places), we follow it.

Engineering precision

For tools used in engineering contexts (pressure, force, fuel economy) we preserve the underlying full-precision value and provide enough significant figures to support downstream calculations. For safety-critical engineering you should still apply your discipline's specific rounding and tolerance conventions.

Errors introduced by rounding

The maximum error introduced by display rounding is less than half of the last displayed digit. Chained conversions performed inside a single tool do not accumulate display-rounding error because intermediate values stay at full precision.