Documentation Portal
Complete operator guides, systems architecture whitepaper, Python SDK integrations, and TOML configuration reference for HyperCache.
Welcome to the HyperCache technical documentation. Choose a section below or use the sidebar to navigate across single-node setup, multi-node cluster coordination, and zero-copy Python integrations.
Operator Quickstart & Cluster Setup
Step-by-step installation, single-node zero-port startup, and multi-node GPU cluster setup with the Central Controller.
Systems Architecture Whitepaper
Deep-dive into raw NVMe auto-discovery, Moka W-TinyLFU metadata, 4 MiB round-robin striping, Direct Block DMA, and async write-back.
Python Data Plane & Framework Integrations
Zero-copy integrations for PyTorch (fsspec), Safetensors, vLLM, Pipelined Layer Loading, PyTorch Distributed Checkpointing (DCP), and KV-Cache prefix blocks.
Configuration Reference (TOML & Environment)
Complete reference for hypercache-agent.toml, hypercache-controller.toml, storage backing modes, and environment variable overrides.
CLI & Fleet Admin Reference
Structured command reference for the node worker daemon (./hypercache) and the cluster fleet coordinator (./hypercache-admin).
Security Model, Network Ports & Prometheus Monitoring
Zero-exposure default security posture, Bearer token authentication, firewall hardening, and Prometheus/Grafana metrics.