shmpipeline

shmpipeline builds local process-based compute pipelines on top of named shared-memory streams from pyshmem.

The documentation site is organized around four user paths:

• getting a pipeline running quickly

• understanding the CLI, runtime model, and GUI

• studying complete worked examples

• looking up the Python API and built-in kernel catalog

Getting Started

• Installation
  • PyPI installs
  • Source installs
  • Platform notes
  • Verify the install
• Quick Start
  • Python API
  • Minimal YAML config
  • CLI flow
  • Next steps
• Configuration Model
  • Shared memory definitions
  • Kernel definitions
  • Auxiliary inputs
  • Graph validation rules
  • CPU and GPU pipelines
  • Custom operations
  • Where to look next

Guides

• CLI Guide
  • Validate
  • Describe
  • Run
  • Serve
  • Logging
  • Recommended workflow
• Control Plane
  • Install
  • Start the service
  • Endpoint surface
  • Minimal Control GUI
  • Python client
  • Live event stream
  • Design constraints
• Runtime And Observability
  • Lifecycle
  • Graph Introspection
  • Runtime snapshots
  • Worker health states
  • Synthetic inputs
  • Viewer behavior
  • Failure handling
• GUI Guide
  • Launch
  • Main areas of the application
  • Typical editing flow
  • Shared memory editor
  • Kernel editor
  • Runtime controls
  • Synthetic inputs
  • Stream viewers
  • Screenshot assets
• Custom Kernel Extensions
  • Example
  • Design guidance
  • Testing guidance
  • Where to look in the API
• Troubleshooting
  • Validation failures before startup
  • Worker stays in waiting-input
  • Worker becomes idle
  • GPU viewer or CPU reader cannot inspect a stream
  • GPU runtime requirements
  • Repeated lifecycle testing

Worked Examples

• Worked Examples
  • Affine Transformation
  • GPU Affine Transformation
  • Custom Operations
  • GPU Custom Operations
  • Basic AO System
  • GPU Basic AO System
  • Observatory AO System
  • How to use the examples

API Reference

• API Reference
  • Core API
  • GUI API
  • Built-In Kernels