Expand description
Burn multi-backend router.
Modules§
- duo
- Module containing the essential types for multi-backend operations.
- quad
- Module containing the essential types for multi-backend operations.
- trio
- Module containing the essential types for multi-backend operations.
Macros§
- binary_
bool_ ops - binary_
float_ cmp_ ops - binary_
float_ ops - binary_
int_ cmp_ ops - binary_
int_ ops - int_
float_ ops - reduce_
float2int_ dim_ ops - reduce_
float_ dim_ ops - reduce_
int_ dim_ ops - scalar_
float2int_ ops - scalar_
float_ cmp_ ops - scalar_
float_ dim_ ops - scalar_
float_ ops - scalar_
int_ cmp_ ops - scalar_
int_ dim_ ops - scalar_
int_ ops - unary_
float_ ops - unary_
int_ ops
Structs§
- Backend
Router - A backend that forwards the tensor operations to the appropriate backend (given multiple backends).
- Byte
Bridge - Simply transfers tensors between backends via the underlying tensor data.
- Direct
Channel - A local channel with direct connection to the backend runner clients.
- Router
Tensor - Tensor primitive for the router backend.
- Runner
- A runner is responsible for executing tensor operations for a given intermediate backend.
- Runner
Context - A runner’s context contains a handle container to manage (i.e., fetch and update) existing tensors.
Traits§
- Multi
Backend Bridge - Allows tensors to be transferred between multiple backends.
- Runner
Channel - Defines the connection channel and operations for a setup with multiple backend runner clients.
- Runner
Client - Define how to interact with the runner.
Functions§
- get_
client - Get the client for the given device
Type Aliases§
- Client
- Type alias for
<R as RunnerChannel>::Client
. - Direct
Byte Channel - A local channel with a simple byte bridge between backends. It transfers tensors between backends via the underlying tensor data.
- Router
- Router backend.
- Tensor
Handle - Type alias for
<Br as MultiBackendBridge>::TensorHandle
.