Module persistent 

Resident persistent-kernel queue engine. Persistent-kernel queue execution for wgpu compute pipelines.

This module provides host-side queue management and GPU dispatch for persistent kernels. A persistent pipeline is compiled once and reused across multiple work items, with only buffer contents changing between calls. If the wgpu device is lost, the backend recovers and the pipeline is recompiled automatically.

Structs

PersistentKernelReport

Resident-kernel execution summary.

PersistentPayloadWorkItem

One unit of persistent-kernel work.

PersistentQueue

GPU-side queue contract for persistent kernels.

WorkResult

Output produced for one persistent-kernel work item.

Functions

run_persistent_kernel

Compile a persistent pipeline and drain queue through the GPU.