ruda-tensor-wgpu 0.21.26

Ruda WGPU tensor device adapter.
Documentation

ruda-tensor-wgpu

Framework tensor adapter over ruda-driver-wgpu. It exposes a Wgpu backend with optional fusion and graphics-API-specific compiler selections.

Interfaces

  • Wgpu<F, I, B> selects floating-point, integer, and Boolean element types.
  • WgpuDevice, runtime options, and initialization helpers are re-exported.
  • With fusion, Wgpu wraps the device backend in ruda_fusion::Fusion; without it, the alias names the device backend directly.

Usage

Cargo package: ruda-tensor-wgpu. Rust import: ruda_tensor_wgpu.

[dependencies]
ruda-tensor-wgpu = "0.21"

Features

Default features: std, autotune, fusion, ruda-tensor-device/default, ruda-core/id-tests-std, ruda-test-runtime/default.

Feature Purpose
fusion Enable tensor-operation fusion.
vulkan Enable the Vulkan/SPIR-V adapter.
webgpu Enable the WebGPU/WGSL adapter.
metal Enable the Metal/MSL adapter.
template Enable source-template kernel integration.

Links