F32

F32 refers to a 32-bit floating-point numeric type commonly used in computing. It is usually synonymous with IEEE 754 single-precision and is also known as float32 or simply float in many programming environments. It is designed to balance range, precision, and memory usage for numeric computations.

A 32-bit float is organized into three fields: 1 sign bit, 8 exponent bits, and 23 fraction (significand) bits. The exponent uses a bias of 127, and normalized numbers have an implicit leading 1 before the binary point. This structure enables a wide range of values, from very small to very large, with approximately seven decimal digits of precision.

Special values and subnormal numbers are defined for edge cases. If the exponent is all zeros and the fraction is nonzero, the value is a subnormal number. If the exponent is all zeros and the fraction is zero, the value is zero. If the exponent is all ones, the value is either infinity (finite exponent) or NaN (not a number) for signaling or quiet NaNs. The finite range is approximately from ±1.17549435 × 10^-38 (minimum normal) up to ±3.4028235 × 10^38, with a precision of about seven decimal digits.

F32 is widely used in performance-sensitive contexts where memory bandwidth and speed are priorities, such as graphics processing, machine learning inference, and scientific computing. It is supported across major CPUs and GPUs and is commonly exposed in high-level languages and libraries as float32, float, or similar typedefs. For comparison, double-precision (float64) offers greater range and accuracy at the cost of memory and computation. See also IEEE 754 and related floating-point formats like float16.