Enum OnnxExecutionProvider

Namespace

VisioForge.Core.Types.X.AI

Assembly

VisioForge.Core.dll

Execution provider used by the ONNX Runtime inference session.

public enum OnnxExecutionProvider

Fields

CPU = 0

CPU execution provider. Always available.

CUDA = 1

NVIDIA CUDA execution provider (Windows/Linux with a compatible CUDA-capable GPU and runtime).

DirectML = 2

DirectML execution provider (Windows, DirectX 12 capable GPU).

CoreML = 3

CoreML execution provider (Apple platforms).

Auto = 4

Automatically selects the fastest hardware provider that is actually present in the loaded ONNX Runtime native build (CUDA, then DirectML, then CoreML), and falls back to VisioForge.Core.Types.X.AI.OnnxExecutionProvider.CPU when no hardware provider is available. This is the default.

Remarks

The selected provider determines which hardware backend runs the model. When a hardware provider is requested but unavailable on the host (missing driver, runtime package, or device), the inference engine logs a warning and falls back to the CPU execution provider, which is always appended. Use VisioForge.Core.Types.X.AI.OnnxExecutionProvider.CPU for the most portable behaviour.