Struct CudaEGraphStructuralCanonicalizationFixedPointReport 

pub struct CudaEGraphStructuralCanonicalizationFixedPointReport {
    pub rounds: Vec<CudaEGraphStructuralCanonicalizationRoundResult>,
    pub final_snapshot: Option<CudaEGraphResidentColumnSnapshot>,
    pub final_signature_snapshot: Option<CudaEGraphResidentSignatureSnapshot>,
    pub final_readback: CudaEGraphFixedPointReadback,
    pub final_full_readback_bytes: usize,
    pub final_signature_snapshot_bytes: usize,
    pub final_additional_readback_bytes: usize,
    pub avoided_final_readback_bytes: usize,
    pub converged: bool,
    pub max_rounds: usize,
    pub total_discovered_pairs: u64,
    pub total_rewrites: u64,
}

Policy-aware result for iterative CUDA-resident e-graph canonicalization.

Fields

rounds: Vec<CudaEGraphStructuralCanonicalizationRoundResult>

Rounds executed, including the final no-op round when convergence is proven before max_rounds.

final_snapshot: Option<CudaEGraphResidentColumnSnapshot>

Full resident columns after the last executed round when requested by CudaEGraphFixedPointReadback::FullColumns.

final_signature_snapshot: Option<CudaEGraphResidentSignatureSnapshot>

Signature-only resident snapshot after the last executed round when requested directly or derivable from a full-column readback.

final_readback: CudaEGraphFixedPointReadback

Final host readback policy used by this run.

final_full_readback_bytes: usize

Bytes a full final resident-column readback would transfer.

final_signature_snapshot_bytes: usize

Bytes represented by the final row-signature column.

final_additional_readback_bytes: usize

Additional bytes transferred at the final readback boundary by the selected policy. Signature-only reports reuse the already-current planning snapshot and therefore do not force another device-to-host copy.

avoided_final_readback_bytes: usize

Full-column final readback bytes avoided by the selected policy.

converged: bool

true iff a no-op discovery round proved fixed-point convergence.

max_rounds: usize

Maximum number of rounds requested by the caller.

total_discovered_pairs: u64

Total unique equivalence pairs discovered across all rounds.

total_rewrites: u64

Total canonical rewrite records applied across all rounds.

Trait Implementations

impl Clone for CudaEGraphStructuralCanonicalizationFixedPointReport

fn clone(&self) -> CudaEGraphStructuralCanonicalizationFixedPointReport

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for CudaEGraphStructuralCanonicalizationFixedPointReport

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

Formats the value using the given formatter.

impl PartialEq for CudaEGraphStructuralCanonicalizationFixedPointReport

fn eq( &self, other: &CudaEGraphStructuralCanonicalizationFixedPointReport, ) -> 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 Eq for CudaEGraphStructuralCanonicalizationFixedPointReport

impl StructuralPartialEq for CudaEGraphStructuralCanonicalizationFixedPointReport