Description
Return binary32 normal-binade/subnormal ULP spacing as exact power-of-two exponent; at powers of two this is the toward-increasing-magnitude binade spacing, not both neighboring gaps.
Binary floating-point storage separates sign, biased exponent and fraction fields. Zero and subnormal patterns use a different significand rule from normal values, and reserved exponent patterns represent infinity or NaN. These tools accept exact unsigned raw patterns as decimal integer text, preserving signed-zero and payload evidence without converting the bits into a JavaScript numeric value.
Return binary32 normal-binade/subnormal ULP spacing as exact power-of-two exponent; at powers of two this is the toward-increasing-magnitude binade spacing, not both neighboring gaps.
When to use this tool
Use this tool to return binary32 normal-binade/subnormal ULP spacing as exact power-of-two exponent; at powers of two this is the toward-increasing-magnitude binade spacing, not both neighboring gaps.
How to use the tool and how it works
- Enter pattern using the input definitions below.
- Run the operation. Resolve malformed-input errors before interpreting the result.
- Read result.binaryExponent. Interpret signs, units, ordering and null values using the rules below.
Read an exact unsigned 32-bit pattern from decimal text. Its fields are one sign bit, 8 exponent bits and 23 fraction bits, with exponent bias 127. Reject infinity and NaN. Return −149 for zero/subnormal patterns or storedExponent−127−23 for normals; the spacing represented is 2^binaryExponent.1, 2
| Input field | Meaning and units | Accepted input |
|---|---|---|
| pattern | Exact unsigned 32-bit raw pattern as decimal integer text, from 0 through 2^32−1. | Text, at most 8,192 UTF-16 code units; the stated grammar and narrower limits also apply. |
| Output field | Interpretation |
|---|---|
| result | Object containing the fields below. |
| result.binaryExponent | Power-of-two exponent: spacing is exactly 2^binaryExponent. |
{
"input": {
"pattern": "1065353216"
},
"output": {
"result": {
"binaryExponent": -23
}
}
}Limitations and assumptions
- At an exact power of two, this is the spacing in its binade toward increasing magnitude. It need not equal both neighboring gaps, and the largest finite value’s next representable value is infinity.
- Pattern width is fixed by the tool name. Supply 0≤pattern<2^width as exact decimal integer text, not hexadecimal text or the represented float’s decimal value. Raw-bit operations preserve payloads without guaranteeing runtime NaN behavior.
- Every declared top-level input is required and unknown fields are rejected. Ordinary strings are limited to 8,192 UTF-16 code units and lists to 128 entries, with tighter operation-specific bounds where stated. JSON text has bounded nesting and node counts.
- Numeric outputs must be finite and within the absolute safe-integer magnitude; exact larger integers use decimal text. Source positions are zero-based UTF-16 offsets and span ends are exclusive unless a different convention is explicitly stated.
Alternative or Complementary methods
Distinguish a raw unsigned bit pattern from the decimal value that pattern represents. Use field join/split operations for exact bit editing and a separate numeric conversion when a runtime floating-point value is required.
References
Similar or alternative tools
- IEEE Binary32 Adjacent Finite Patterns
Return adjacent finite binary32 raw value patterns toward positive/negative infinity; signed zeros skip the equivalent other zero and extrema return null instead of infinity.
- IEEE Binary32 Exact Dyadic Record
Decode an exact IEEE binary32 raw pattern into signed integer significand times 2^binaryExponent; infinity/NaN have null dyadic value, negative zero remains explicit.