shader-edsl-compute

What problem does it solve?

Rendiation’s shader EDSL compute reference helps you design correct compute dispatches by explaining how to build compute pipelines and write GPU compute logic (IDs, memory spaces, synchronization, and common reduction patterns).

Core Features & Use Cases

• Compute IDs & dispatch mapping: Use global/local/workgroup and subgroup IDs for indexing and workload decomposition.
• Synchronization primitives: Apply storage, workgroup, and subgroup barriers to make memory interactions correct and deterministic.
• Workgroup memory patterns: Define workgroup shared variables (including host-sized arrays) and use workgroup-uniform loads.
• GPU-focused recipes: Implement workgroup reduction, per-invocation counting loops, storage buffer iteration, SSAO-style accumulate flows, and subgroup prefix sums.
• GPU unit testing workflow: Run compute passes end-to-end (GPU init → dispatch → readback) to validate shader behavior.

Quick Start

Use shader-edsl-compute to implement and test a compute pipeline that writes results into a storage buffer using correct invocation indexing and optional GPU unit tests.