SOKR — Sovereign Open Kernel Runtime

Immutable sovereign core. Everything else a plugin.

This crate is in early design phase. No API is stable.

What is SOKR?

SOKR is a sovereign compute runtime written in Rust where the core is immutable and everything else is a plugin: IR representations, substrate backends, language bindings, and dispatch policy.

The core exposes exactly three operations across a stable C ABI surface:

Operation	Question
Capability	Can this substrate fulfill this computation?
Dispatch	Fulfill it.
Completion	Signal when fulfilled.

No assumption is made about memory model, parallelism, execution time, or computation representation. Any substrate that can answer three questions is a valid SOKR backend — including substrates that do not yet exist.

Philosophy

The algorithm is the permanent asset. The substrate is a runtime decision.

SOKR imposes no conditions on the algorithm, the substrate, or the user. No vendor can revoke it. No export restriction can strand it. No hardware generation can obsolete it.

Architecture

User code (any language)
        ↓
IR Plugin          ← swappable: SOKR-native, SPIR-V, OpenQASM, or future
        ↓
SOKR Core          ← immutable, no_std, C ABI
  Capability → Dispatch → Completion
        ↓
Substrate Plugin   ← swappable: GPU, CPU, QPU, Neuromorphic, Photonic, or future
        ↓
Hardware

Everything above and below the core is a plugin. The core contains no assumptions — not about computation representation, not about hardware model, not about language, not about security policy.

Target Substrates

Current and future:

GPU — NVIDIA (PTX/CUDA), AMD (HIP), Intel Arc (SPIR-V)
CPU — fallback, always available
QPU — quantum processors (OpenQASM 3)
Neuromorphic — Intel Loihi, IBM NorthPole
Photonic — Lightmatter and successors
WebGPU — browser and edge compute

Status

v0.1.0 — name reservation. Design phase. No runnable code.

Roadmap and architecture forthcoming at: https://github.com/sokr-rs/sokr

License

Licensed under either of

MIT license (LICENSE-MIT)
Apache License, Version 2.0 (LICENSE-APACHE)

at your discretion.

Any contribution intentionally submitted for inclusion in this work shall be dual licensed as above, without any additional terms or conditions.