vector.fptrunc¶
Lanewise floating-point precision truncation using round-to-nearest, ties-to-even. Source and result shapes match exactly; only the floating-point element type narrows. Special values follow the destination format: f8E4M3 saturates finite overflow and infinities to its signed maximum finite value while preserving NaNs; IEEE formats preserve infinities and NaNs. Without flags, the conversion preserves subnormal inputs and results. daz permits replacing a subnormal input with its signed zero before conversion; ftz permits replacing a subnormal result with its signed zero.
Operation contract¶
| Property | Value |
|---|---|
| Semantic phase | executable |
| Traits | Pure, Elementwise, Decomposable, SafeToSpeculate |
Signature¶
| Kind | Name | Type | Cardinality | Description |
|---|---|---|---|---|
| Operand | input |
vector |
required | — |
| Result | result |
vector |
required | — |
| Attribute | subnormal |
flags FloatConversionFlags |
optional | — |
Verification constraints¶
HasFloatElement(input)HasFloatElement(result)SameKind(input, result)SameShape(input, result)ElementWidthLessThan(result, input)
Examples¶
Note
A canonical source example has not been added for this operation yet.