Skip to content
Oguzhan KAYNAK edited this page Nov 14, 2026 · 2 revisions

gridlane

A layered, loosely coupled model for data distribution and messaging with low latency, between processes and across machines.

gridlane is a design, not a product. It shows how a few small, independent layers can be stacked so that:

  • an application reads its data from local memory, as if it had shared memory, whatever connects the machines;
  • the same binaries run over TCP across machines, or over shared memory on one machine, with only a configuration file changed;
  • every layer can be extended with a new API (a transport, a device, a serializer) without touching the others.

The focus is data distribution for parallel algorithms. The same layers also let any two independent applications communicate.

The layers in one picture

flowchart TB
  app["your application"]
  subgraph gridlane["gridlane: composition"]
    direction LR
    prim["data distribution primitives<br/>put · signal · wait · collectives"]
    glue["adapters"]
  end
  wf["wireforge<br/>typed messages, generated codecs"]
  pr["pairring<br/>message channel"]
  ps["pushseg<br/>segments: write far, read local"]
  tr["transports<br/>TCP · POSIX shared memory · your API"]

  app --> prim
  app --> wf
  wf -->|control plane| pr
  pr --> ps
  prim -->|data plane| ps
  ps --> tr
Loading

Pages

Page What it explains
Layered Model Each layer, its contract, and why the layers do not know each other
Programming Model Targets, applications, workers, ranks; symmetric memory; put / signal / wait; collectives
Portability The same run over TCP on N machines and over shared memory on one machine
Extending Adding a new API at each layer
Demonstrator The Gray-Scott pipeline: both planes in one program
Getting Started Build, test, run

Repositories

Repository Role
gridlane Composition, runtime, demonstrator
pushseg Segments and the memory contract
pairring Point-to-point message channel
wireforge Message definitions and code generation

The full design document is docs/architecture.md.

Clone this wiki locally