kernel.subgroup.reduce

← kernel dialect

Reduce a scalar or rank-1 vector value across the current subgroup. Each active contribution participates once per component; association is unspecified. Floating sums and products default to native target arithmetic, which may change signed zeros, flush subnormals, or relax nonfinite handling. The optional preserve requirement instead evaluates each combine with the source format's arithmetic semantics. Neither mode promises serial-reference bits, reproducibility, or a correctly rounded real-number sum. Integer arithmetic and the selected floating-extremum kind keep their exact semantics.

Operation contract

Property	Value
Semantic phase	executable
Traits	Convergent, HasAnyAncestor(kernel.def, pipeline.strand)
Target contracts	kernel.synchronization

Signature

Kind	Name	Type	Cardinality	Description
Operand	value	any	required	Per-invocation value to reduce.
Result	result	any	required	Reduced value.
Attribute	kind	enum CombiningKind	required	Combining operation.
Attribute	math	flags CollectiveMathFlags	optional	Optional floating-point preservation requirement; omission selects native arithmetic.
Attribute	cluster_size	i64	optional	Optional clustered subgroup size.
Attribute	cluster_stride	i64	optional	Optional clustered subgroup stride.

Verification constraints

• SameType(value, result)

Examples

%sum = kernel.subgroup.reduce<addf> %v : f32