Language

ArmAIInfraOptimizationModelFactory Class

Definition

A factory class for creating instances of the models for mocking.

public static class ArmAIInfraOptimizationModelFactory
type ArmAIInfraOptimizationModelFactory = class
Public Class ArmAIInfraOptimizationModelFactory
Inheritance
ArmAIInfraOptimizationModelFactory

Methods

Name Description
AIInfraOptimizationRecommendVmSkuOptionConfig(String, Int32, RecommendVmSkuReadiness, RecommendVmSkuTopology, RecommendVmSkuCost, RecommendVmSkuQuota, RecommendVmSkuPlacement, Nullable<Boolean>, RecommendVmSkuBenchmark, Uri, RecommendVmSkuPerformanceEstimate)

One ranked virtual machine size recommendation.

RecommendVmSkuAccelerator(String, Int32, Int32)

The accelerator hardware available in each virtual machine.

RecommendVmSkuBenchmark(RecommendVmSkuBenchmarkLevel, String, String, Int32, String, String, String, String, Nullable<Double>, String, String, String, String, Nullable<Double>, Nullable<Double>)

The external benchmark evidence selected for one virtual machine size option.

RecommendVmSkuContent(RecommendVmSkuModel, RecommendVmSkuWorkload, RecommendVmSkuDeployment, RecommendVmSkuOptimization, IEnumerable<String>, RecommendVmSkuSubscriptionConfig)

The parameters used to request an Azure virtual machine size recommendation for an AI workload.

RecommendVmSkuCost(String, Double, Double, RecommendVmSkuCostSource, Nullable<DateTimeOffset>, RecommendVmSkuStageError)

The deployment cost for one virtual machine size option.

RecommendVmSkuDeployment(Nullable<Int32>, Nullable<RecommendVmSkuProcurementOption>)

Deployment scale and procurement inputs.

RecommendVmSkuHourlyCost(Nullable<Double>, String)

A currency-qualified deployment-wide hourly cost.

RecommendVmSkuMemory(Double, Double, Double, Double)

The memory calculation shared by every returned option.

RecommendVmSkuModel(String, Nullable<Int64>, Nullable<WeightPrecision>, Nullable<Int32>, Nullable<Int32>, Nullable<Int32>, Nullable<Int32>)

Model inputs used for metadata resolution and memory sizing.

RecommendVmSkuOptimization(Nullable<RecommendVmSkuRankingPreference>, RecommendVmSkuHourlyCost)

Ranking preference and optional deployment-wide hourly cost limit.

RecommendVmSkuPerformanceEstimate(Double, Double)

An advisory deployment throughput and per-request latency estimate.

RecommendVmSkuPlacement(RecommendVmSkuPlacementStatus, String, RecommendVmSkuStageError, IEnumerable<String>)

The capacity placement evidence for one virtual machine size option.

RecommendVmSkuQuota(RecommendVmSkuQuotaStatus, String, RecommendVmSkuStageError, Nullable<Int32>, RecommendVmSkuQuotaIncrease)

The subscription quota evidence for one virtual machine size option.

RecommendVmSkuQuotaIncrease(IEnumerable<RecommendVmSkuQuotaLimit>, Uri)

The quota limits to raise before retrying a deployment.

RecommendVmSkuQuotaLimit(String, Int64, Int64, Int64)

One quota limit that blocks the requested deployment.

RecommendVmSkuReadiness(RecommendVmSkuReadinessStatus, String)

How ready a recommended option is to deploy.

RecommendVmSkuResult(IEnumerable<AIInfraOptimizationRecommendVmSkuOptionConfig>, RecommendVmSkuSizing)

The recommended Azure virtual machine sizes for the workload, with the sizing calculation they were derived from.

RecommendVmSkuSizing(RecommendVmSkuSizingModel, RecommendVmSkuSizingWorkload, RecommendVmSkuMemory, RecommendVmSkuSizingBasis)

The resolved workload sizing shared by every recommended option.

RecommendVmSkuSizingBasis(ParameterCountSource, ModelWeightsCalculation, ContextLengthSource)

The sources used for the sizing calculation.

RecommendVmSkuSizingModel(Int64, WeightPrecision, String)

The resolved model identity and weight precision.

RecommendVmSkuSizingWorkload(Int32, Int32)

The resolved per-replica workload inputs used for memory sizing.

RecommendVmSkuStageError(RecommendVmSkuStageErrorCode, String)

An error from an optional subscription-scoped recommendation stage.

RecommendVmSkuSubscriptionConfig(Nullable<Boolean>, Nullable<Boolean>, Nullable<Boolean>)

The subscription-scoped lookups to include in the recommendation.

RecommendVmSkuTokenBudget(Nullable<Int32>, Nullable<Int32>)

The per-request token budget used to size the inference key-value cache.

RecommendVmSkuTopology(Int32, Int64, RecommendVmSkuAccelerator)

The deployment topology for one virtual machine size option.

RecommendVmSkuWorkload(Nullable<Int32>, RecommendVmSkuTokenBudget)

Inference workload assumptions that affect memory sizing.

Applies to