Plain-AyI
← All projects
Running AI at scale

NCCL

NVIDIA's library that lets many graphics cards share data quickly, used when training large AI models.

New this week

What it does

NCCL (said "Nickel") is a library from NVIDIA for communication between graphics cards (GPUs). When large AI models are trained on many GPUs at once, the cards constantly need to swap and combine results, and NCCL handles that.

It is tuned for fast connections inside a machine and across networks between machines.

Official tested builds can be downloaded from NVIDIA, or developers can build it from this source code.

Who it's for

Engineers building or running AI training on many NVIDIA GPUs.

Can I use it without coding?

Developers only. It is low-level software for systems engineers. New to this? Start here.

Can I run it?

Cost
FreeNothing to pay or sign up for.
Runs on
Linux, Server
Computer
Made for servers or data centres
Privacy
Runs on your own device: your data stays with you
API key
Not needed

Try it

Using Claude Code, Cursor or another AI coding assistant? Paste the prompt and it will walk you through the setup.New to this?

Before you install

  • Only download from the official GitHub page. Copies with similar names exist for many popular projects.
  • No standard license listed. Check the GitHub page before using it for business.
  • More on staying safe

Where people are talking about it