cxxnet

Health v2 · maintenance only

Flags: archived no_license

Adoption not part of the score

1024 stars · 406 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

ps-lite is a lightweight, efficient C++ implementation of the parameter server framework for distributed machine learning. It exposes simple Push/Pull/Wait APIs for key-value communication between worker and server nodes, with zero-copy transfers and user-defined server-side handles.

Use cases

• synchronize model parameters across distributed training workers
• build distributed deep learning systems like MXNet or BytePS
• run distributed sparse logistic regression or factorization machines
• push gradients to server nodes and pull updated weights
• compress communication with user-defined filters
• implement custom server-side update logic for KV stores

When to choose

• you need a minimal, embeddable parameter server layer in C++ for distributed ML
• you are building or extending distributed training frameworks that communicate via key-value push/pull
• you need high-performance, zero-copy asynchronous communication between workers and servers

When to avoid

• you want a full-featured distributed training framework rather than a communication layer
• your project is Python-only and you prefer higher-level APIs like those in MXNet or PyTorch DDP
• you need actively developed features; the project is largely in maintenance mode

Facets

library · maturity maintenance

machine-learning deep-learning llm-training networking concurrency machine-learning deep-learning microservices large-language-models cpp cross-platform parameter-server distributed-training kvstore zero-copy async-communication dmlc linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="dmlc/cxxnet")