# NVIDIA/nccl

> NCCL is a standalone C++ library of collective communication primitives (all-reduce, all-gather, broadcast, reduce-scatter, send/receive) optimized for high-bandwidth inter-GPU communication over PCIe, NVLink, NVSwitch, InfiniBand and TCP/IP.

- Magnitude: 0.5 out of 10 — Quiet
- Stars: 5,178 total · +0 stars measured 2026-10-22, 10:42–13:03 UTC, ≈ 0 by evening
- Star trust: star growth looks organic
- Category: DevOps & infrastructure · Language: C++ · Created: 2015-11-14 · Last push: 2026-10-21
- GitHub: https://github.com/NVIDIA/nccl · Homepage: https://docs.nvidia.com/deeplearning/nccl/user-guide/docs/index.html · Page: https://gitnova.dev/en/r/NVIDIA/nccl

## Useful for

- Run all-reduce across multiple GPUs in a single node or cluster for distributed training
- Build NCCL as a deb, rpm, txz package or Python wheel for deployment
- Benchmark collective bandwidth with nccl-tests on your GPU topology

## Why it’s here

- Star-counter measurements on 2026-10-22 (UTC), 10:42–13:03: 5178 → 5178 stars (+0). This is the change over that interval.
- GitHub Trending C++ today: #8.
- Recent forks include notable developers: @jeffhammond (840 followers).

## Star trust

Star growth looks organic.

Star-trust labels are heuristics based on the repository’s behavior, not a check of every stargazer.

## Numbers

- Forks: 1,450
- Issues and pull requests: 2,470
- Watchers: 164
- Average over the last week: 2 per day
- Usual pace: 3 per day
- Stars in the last hour (measured): 0
- Latest release: v2.32.3-1 (2026-09-17)

## Stars per day, last 30 days (oldest → newest, today is partial)

2026-09-23 … 2026-10-22: 2, 3, 2, 3, 1, 2, 5, 2, 2, 2, 2, 2, 1, 5, 2, 3, 4, 3, 1, 6, 5, 1, 3, 2, 3, 2, 0, 5, 0, 0

## Spotted in now

- GitHub Trending C++ today: #8

## More in this category

1. **perplexityai/SPACE** — 4.1 · Cooling · DevOps & infrastructure · Rust · +51 stars measured 2026-10-22, 00:38–12:53 UTC, ≈ 85 by evening
   A platform for running code in isolated Firecracker microVMs. Applications use its HTTP API to create Linux environments, execute commands, and manage sandbox lifecycles.
   Full card: https://gitnova.dev/en/r/perplexityai/SPACE.md
2. **Samsara-1999/Bursar** — 3.8 · Early signal · DevOps & infrastructure · Go · +0 stars measured 2026-10-22, 00:04–12:54 UTC, ≈ 20 by evening
   Self-hosted platform for governing shared Linux GPU servers: central accounting of GPU/CPU usage, quotas, points, account mapping and access policies, while users keep their normal SSH workflow.
   Full card: https://gitnova.dev/en/r/Samsara-1999/Bursar.md
3. **ChiefGyk3D/Clouds-Against-Humanity** — 3.7 · Breakout · DevOps & infrastructure · Python · +95 stars measured 2026-10-22, 00:39–12:54 UTC, ≈ 167 by evening
   A teaching cloud lab on DigitalOcean with zero-trust access: OpenTofu stacks applied via Spacelift, secrets in Doppler, reachable only through Twingate, with no public inbound ports.
   Full card: https://gitnova.dev/en/r/ChiefGyk3D/Clouds-Against-Humanity.md
4. **rohit267/net-merge** — 3.4 · Peaking · DevOps & infrastructure · Go · +102 stars measured 2026-10-22, 00:40–12:54 UTC, ≈ 162 by evening
   A cross-platform tool that aggregates multiple internet connections at once, load-balancing each new connection across Wi-Fi, Ethernet and tethers. It is a free alternative to Speedify, without any account, server or VPN.
   Full card: https://gitnova.dev/en/r/rohit267/net-merge.md
5. **kubernetes/kubernetes** — 3.3 · Steady · DevOps & infrastructure · Go · +37 stars measured 2026-10-22, 00:39–12:55 UTC, ≈ 62 by evening
   Kubernetes is an open source container orchestration system for deploying, scaling, and managing containerized applications across multiple hosts.
   Full card: https://gitnova.dev/en/r/kubernetes/kubernetes.md

---

Magnitude (0–10) measures how fast and how unusually interest in a repository is growing right now. It is not a quality score.
Days are UTC. “So far today” is a fact; “expected by the end of the day” is a forecast.
Summaries and use cases are written by an LLM (DeepSeek V4.1 Flash) from the README and may be inaccurate: verify specific claims (benchmarks, speed, hardware) in the repository itself.
Data as of 2026-10-22 13:05 UTC, updated every 30 minutes.