Struct ort::execution_providers::ExecutionProvider

pub struct ExecutionProvider { /* private fields */ }

Execution provider container. See the ONNX Runtime docs for more info on execution providers. Execution providers are actually registered via the with_execution_providers() functions crate::SessionBuilder (per-session) or EnvBuilder (default for all sessions in an environment).

Implementations

impl ExecutionProvider

pub fn new(provider: impl Into<String>) -> Self

Creates an ExecutionProvider for the given execution provider name.

You probably want the dedicated methods instead, e.g. ExecutionProvider::cuda.

pub fn cpu() -> Self

Creates a new CPUExecutionProvider configuration object.

pub fn cuda() -> Self

Creates a new CUDAExecutionProvider configuration object.

pub fn tensorrt() -> Self

Creates a new TensorrtExecutionProvider configuration object.

pub fn acl() -> Self

Creates a new AclExecutionProvider configuration object.

pub fn dnnl() -> Self

Creates a new DnnlExecutionProvider configuration object.

pub fn onednn() -> Self

Creates a new DnnlExecutionProvider configuration object.

pub fn coreml() -> Self

Creates a new CoreMLExecutionProvider configuration object.

pub fn directml() -> Self

Creates a new DmlExecutionProvider configuration object.

pub fn rocm() -> Self

Creates a new ROCmExecutionProvider configuration object.

pub fn is_available(&self) -> bool

Returns true if this execution provider is available, false otherwise. The CPU execution provider will always be available.

pub fn with(self, name: impl Into<String>, value: impl Into<String>) -> Self

Configure this execution provider with the given option name and value

pub fn with_use_arena(self, v: bool) -> Self

Whether or not to use the arena allocator.

Supported backends: CPU, ACL, oneDNN

pub fn with_device_id(self, v: i32) -> Self

The device ID to initialize the execution provider on.

Supported backends: DirectML

pub fn with_ane_only(self, v: bool) -> Self

By default, the CoreML EP will be enabled for all compatible Apple devices. Setting this option will only enable CoreML EP for Apple devices with a compatible Apple Neural Engine (ANE).

Note: Enabling this option does not guarantee the entire model to be executed using ANE only.

Supported backends: CoreML

Trait Implementations

impl Clone for ExecutionProvider

fn clone(&self) -> ExecutionProvider

Returns a copy of the value.

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

Performs copy-assignment from source.

impl Debug for ExecutionProvider

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

Formats the value using the given formatter.