Struct GraphIr 

pub struct GraphIr {
    pub operations: Vec<OperationIr>,
    pub inputs: Vec<TensorId>,
    pub outputs: Vec<TensorId>,
}

An ordered operation graph with an explicit tensor boundary.

GraphIr::new infers the dependency boundary from the operations. Producers that expose a narrower logical boundary, such as graph capture, may provide explicit inputs and outputs after validating them against GraphIr::classify.

Fields

operations: Vec<OperationIr>

Operations in execution order.

inputs: Vec<TensorId>

Ordered tensor IDs exposed as graph inputs.

outputs: Vec<TensorId>

Ordered tensor IDs exposed as graph outputs.

Implementations

impl GraphIr

pub fn new(operations: Vec<OperationIr>) -> GraphIr

Normalize an operation sequence and classify its tensor boundary in first-use order.

Inputs are tensors that aren’t produced by a compute operation: external data and results from an earlier graph, including outputs of OperationIr::Init. Initializer handles are registered out of band, so they remain graph inputs even though an initialization operation produces them.

Outputs are tensors produced by compute operations that survive the graph. Tensors consumed in place (ReadWrite) or explicitly dropped within the graph are excluded.

Intermediate tensors that are produced and then consumed in place or explicitly dropped appear in neither boundary. Input and output identifiers retain their first-use and production order, respectively, making construction deterministic.

pub fn classify(operations: &[OperationIr]) -> GraphBoundary

Classify the dependency boundary of an operation sequence without cloning it.

This applies the same rules as GraphIr::new and is useful to validate an explicitly declared logical boundary or to estimate graph binding costs.

pub fn len(&self) -> usize

Number of operations in the graph.

pub fn is_empty(&self) -> bool

Whether the graph contains no operations.

Trait Implementations

impl Clone for GraphIr

fn clone(&self) -> GraphIr

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for GraphIr

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for GraphIr

fn deserialize<__D>( __deserializer: __D, ) -> Result<GraphIr, <__D as Deserializer<'de>>::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl From<Vec<OperationIr>> for GraphIr

fn from(operations: Vec<OperationIr>) -> GraphIr

Converts to this type from the input type.

impl PartialEq for GraphIr

fn eq(&self, other: &GraphIr) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for GraphIr

fn serialize<__S>( &self, __serializer: __S, ) -> Result<<__S as Serializer>::Ok, <__S as Serializer>::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for GraphIr