Dynamic range
Beginner
The span from the smallest to the largest number a format can store.
Novice
The ratio between the largest and smallest non-zero values a format can represent, often counted in powers of two (binades). FP32 and BF16 cover about to ; FP16 only about to (down to with subnormals).
Expert
Usually quoted in binades (octaves) from the smallest subnormal to the largest normal: 18 for OCP E4M3 and 32 for E5M2. It decides whether a tensor fits without a scale factor; a scale only shifts the window, it doesn’t widen it.
Explained in Number formats (Architectures).
See also: Exponent (floating point), Scale factor.