BackendPerfContextData class

Native performance timings reported by llama.cpp for the active context.

Constructors

BackendPerfContextData({required double loadMs, required double promptEvalMs, required double evalMs, required double sampleMs, double? decodeMs, required int promptEvalTokens, required int evalTokens, required int sampleCount, required int reusedGraphs, int? speculativeDraftTokens, int? speculativeAcceptedDraftTokens, int? speculativeDraftAttempts, int? speculativeVerifyTokens, int? speculativeReplayTokens, double? speculativeDraftMs, double? speculativeVerifyMs})
Creates a new BackendPerfContextData.
const

Properties

decodeMs → double?
Time spent decoding generated tokens in ms, excluding prompt ingestion.
final
evalMs → double
Time spent generating tokens in ms.
final
evalTokens → int
Number of generated tokens evaluated.
final
hashCode → int
The hash code for this object.
no setterinherited
loadMs → double
Time spent loading the model in ms.
final
promptEvalMs → double
Time spent evaluating prompt tokens in ms.
final
promptEvalTokens → int
Number of prompt tokens evaluated.
final
reusedGraphs → int
Number of times compute graphs were reused.
final
runtimeType → Type
A representation of the runtime type of the object.
no setterinherited
sampleCount → int
Number of sampler steps recorded.
final
sampleMs → double
Time spent sampling generated tokens in ms.
final
speculativeAcceptanceRate → double?
Accepted speculative draft-token ratio, when available.
no setter
speculativeAcceptedDraftTokens → int?
Number of speculative draft tokens accepted by the backend.
final
speculativeDraftAttempts → int?
Number of speculative draft attempts made by the backend.
final
speculativeDraftMs → double?
Time spent generating speculative draft tokens in ms.
final
speculativeDraftTokens → int?
Number of speculative draft tokens proposed by the backend.
final
speculativeReplayTokens → int?
Number of target-model tokens decoded to replay accepted speculative work.
final
speculativeVerifyMs → double?
Time spent verifying speculative draft tokens in ms.
final
speculativeVerifyTokens → int?
Number of target-model tokens decoded to verify speculative drafts.
final

Methods

noSuchMethod(Invocation invocation) → dynamic
Invoked when a nonexistent method or property is accessed.
inherited
toString() → String
A string representation of this object.
inherited

Operators

operator ==(Object other) → bool
The equality operator.
inherited