ruda-tensor-host
CPU tensor backend implemented with the Ruda host-domain libraries. It implements framework tensor operations using host storage, strided layouts, optional SIMD, and optional Rayon parallelism.
Interfaces
HostandHostDeviceare the framework backend and device.HostTensor,HostQTensor, andLayoutrepresent host tensors and layouts.- Use
Hostas the backend parameter ofruda_tensor::api::Tensor; enable the tensor package'sapifeature in the application.
Usage
Cargo package: ruda-tensor-host. Rust import: ruda_tensor_host.
[]
= "0.21"
Features
Default features: std, simd, rayon.
| Feature | Purpose |
|---|---|
simd |
Enable SIMD implementations. |
rayon |
Enable Rayon parallel execution. |
sparse |
Enable host sparse operations. |
critical-section |
Support targets that require critical-section-backed synchronization. |