v1.0.1

Loom is a data-parallelism framework for Zig.

It is lightweight and makes it easy to convert sequential loops into parallel pipelines. Its structured APIs and lock-free internals encourage race-free patterns by design.

Loom is a high-performance work-stealing thread pool with Rayon-style parallel iterators for Zig.

Feature-focused:

Transform sequential loops into parallel pipelines with zero boilerplate—just .map(), .filter(), .reduce().

Performance-focused:

Achieve multi-core throughput with automatic work distribution, lock-free deques, and cache-optimized scheduling.

Developer-focused:

Write par_iter(data).map(transform).reduce(combine) and let Loom handle thread management, load balancing, and work stealing.

Complete pitch:

Loom brings Rayon-style ergonomics to Zig: a work-stealing thread pool, composable parallel iterators, and automatic chunking—turning for loops into multi-core pipelines with a single import

v0.1.5

Zig‑Beam v0.1.5

Date: 2025-11-22

This is a draft release generated from tag v0.1.5.

Quick Start (pin by tag)

In build.zig:

In code:

v0.1.0

Zig‑Beam v0.1.0

Date: 2025-11-21

This is a draft release generated from tag v0.1.0.

Quick Start (pin by tag)

In build.zig:

In code:

v0.1.0-rc3

Zig‑Beam v0.1.0-rc3

Date: 2025-11-12

This is a draft release generated from tag v0.1.0-rc3.

Reports

• Arc benchmarks: docs/utils/arc_benchmark_results.md
• Thread‑Local Cache benchmarks: docs/utils/thread_local_cache_benchmark_results.md
• Dependency graph: docs/utils/dependency_graph.md

Quick Start (pin by tag)

In build.zig:

In code: