Home

Welcome to the cachex Wiki

cachex is a high-performance, concurrent, and "intelligent" in-memory cache for Go 1.23+. It is designed for applications where latency is measured in nanoseconds and hit-rate efficiency is non-negotiable.

🎯 The Three Pillars of cachex

1. Extreme Concurrency: Leveraging a 256-shard architecture to eliminate mutex contention on high-core-count servers.
2. Superior Intelligence: Implementing W-TinyLFU for smart admission and eviction, ensuring your cache contains the hottest data, not just the most recent.
3. Hybrid Flexibility: A seamless L1 (Local) and L2 (Distributed/Redis) store strategy that works out-of-the-box.

🗺️ Navigation

• Architecture Deep Dive: Learn how we achieved 125ns latency.
• W-TinyLFU & Eviction: Understanding the admission policy.
• Hybrid Store Strategy: How L1/L2 promotion works.
• Benchmarks: Reproducing our performance results.
• Roadmap: Where we are heading next.