nixl-cu13 0.9.0

NIXL Python API

NVIDIA Inference Xfer Library (NIXL)

NVIDIA Inference Xfer Library (NIXL) is targeted for accelerating point to point communications in AI inference frameworks such as NVIDIA Dynamo, while providing an abstraction over various types of memory (e.g., CPU and GPU) and storage (e.g., file, block and object store) through a modular plug-in architecture.

Pre-build Distributions

PyPI Wheel

The nixl python API and libraries, including UCX, are available directly through PyPI.

It can be installed for CUDA 12 with:

pip install nixl[cu12]

For CUDA 13 with:

pip install nixl[cu13]

For backwards compatibility, pip install nixl installs automatically nixl[cu12], continuing to work seamlessly for CUDA 12 users without requiring changes to downstream project dependencies.

If both nixl-cu12 and nixl-cu13 are installed at the same time in an environment, nixl-cu13 takes precedence.

Prerequisites for source build

Ubuntu:

$ sudo apt install build-essential cmake pkg-config

Fedora:

$ sudo dnf install gcc-c++ cmake pkg-config

Python

$ pip3 install meson ninja pybind11 tomlkit

UCX

NIXL was tested with UCX version 1.20.x.

GDRCopy is available on Github and is necessary for maximum performance, but UCX and NIXL will work without it.

$ git clone https://github.com/openucx/ucx.git
$ cd ucx
$ git checkout v1.20.x
$ ./autogen.sh
$ ./contrib/configure-release-mt       \
    --enable-shared                    \
    --disable-static                   \
    --disable-doxygen-doc              \
    --enable-optimizations             \
    --enable-cma                       \
    --enable-devel-headers             \
    --with-cuda=<cuda install>         \
    --with-verbs                       \
    --with-dm                          \
    --with-gdrcopy=<gdrcopy install>
$ make -j
$ make -j install-strip
$ ldconfig

ETCD (Optional)

NIXL can use ETCD for metadata distribution and coordination between nodes in distributed environments. To use ETCD with NIXL:

ETCD Server and Client

$ sudo apt install etcd etcd-server etcd-client

# Or use Docker
$ docker run -d -p 2379:2379 quay.io/coreos/etcd:v3.5.1

ETCD CPP API

Installed from https://github.com/etcd-cpp-apiv3/etcd-cpp-apiv3

$ sudo apt install libgrpc-dev libgrpc++-dev libprotobuf-dev protobuf-compiler-grpc
$ sudo apt install libcpprest-dev
$ git clone https://github.com/etcd-cpp-apiv3/etcd-cpp-apiv3.git
$ cd etcd-cpp-apiv3
$ mkdir build && cd build
$ cmake ..
$ make -j$(nproc) && make install

Additional plugins

Some plugins may have additional build requirements, see them here:

• Mooncake
• POSIX
• GDS

Getting started

Build & install

$ meson setup <name_of_build_dir>
$ cd <name_of_build_dir>
$ ninja
$ ninja install

Build Options

Release build (default)

$ meson setup <name_of_build_dir>

Debug build

$ meson setup <name_of_build_dir> --buildtype=debug

NIXL-specific build options

# Example with custom options
$ meson setup <name_of_build_dir> \
    -Dbuild_docs=true \           # Build Doxygen documentation
    -Ducx_path=/path/to/ucx \     # Custom UCX installation path
    -Dinstall_headers=true \      # Install development headers
    -Ddisable_gds_backend=false   # Enable GDS backend

Common build options:

• build_docs: Build Doxygen documentation (default: false)
• ucx_path: Path to UCX installation (default: system path)
• install_headers: Install development headers (default: true)
• disable_gds_backend: Disable GDS backend (default: false)
• cudapath_inc, cudapath_lib: Custom CUDA paths
• static_plugins: Comma-separated list of plugins to build statically
• enable_plugins: Comma-separated list of plugins to build (e.g. -Denable_plugins=UCX,POSIX). Cannot be used with disable_plugins.
• disable_plugins: Comma-separated list of plugins to exclude (e.g. -Ddisable_plugins=GDS). Cannot be used with enable_plugins.

Environment Variables

There are a few environment variables that can be set to configure the build:

• NIXL_NO_STUBS_FALLBACK: If not set or 0, build NIXL stub library if the library build fails

Building Documentation

If you have Doxygen installed, you can build the documentation:

# Configure with documentation enabled
$ meson setup <name_of_build_dir> -Dbuild_docs=true
$ cd <name_of_build_dir>
$ ninja

# Documentation will be generated in <name_of_build_dir>/html
# After installation (ninja install), documentation will be available in <prefix>/share/doc/nixl/

Python Interface

NIXL provides Python bindings through pybind11. For detailed Python API documentation, see docs/python_api.md.

The preferred way to install the Python bindings is through pip from PyPI:

pip install nixl[cu12]

Or for CUDA 13 with:

pip install nixl[cu13]

To build and install the Python bindings from source, you have to build and install separately the platform-specific package and the nixl meta-package:

On CUDA 12:

pip install .
meson setup build
ninja -C build
pip install build/src/bindings/python/nixl-meta/nixl-*-py3-none-any.whl

On CUDA 13:

pip install .
./contrib/tomlutil.py --wheel-name nixl-cu13 pyproject.toml
meson setup build
ninja -C build
pip install build/src/bindings/python/nixl-meta/nixl-*-py3-none-any.whl

For Python examples, see examples/python/.

Rust Bindings

Build

• Use -Drust=true meson option to build rust bindings.
• Use --buildtype=debug for a debug build (default is release).
• Or build manually:

  $ cargo build --release
  

Install

The bindings will be installed under nixl-sys in the configured installation prefix. Can be done using ninja, from project build directory:

$ ninja install

Test

# Rust bindings tests
$ cargo test

Use in your project by adding to Cargo.toml:

[dependencies]
nixl-sys = { path = "path/to/nixl/bindings/rust" }

Other build options

See contrib/README.md for more build options.

Building Docker container

To build the docker container, first clone the current repository. Also make sure you are able to pull docker images to your machine before attempting to build the container.

Run the following from the root folder of the cloned NIXL repository:

$ ./contrib/build-container.sh

By default, the container is built with Ubuntu 24.04. To build a container for Ubuntu 22.04 use the --os option as follows:

$ ./contrib/build-container.sh --os ubuntu22

To see all the options supported by the container use:

$ ./contrib/build-container.sh -h

The container also includes a prebuilt python wheel in /workspace/dist if required for installing/distributing. Also, the wheel can be built with a separate script (see below).

Building the python wheel

The contrib folder also includes a script to build the python wheel with the UCX dependencies. Note, that UCX and other NIXL dependencies are required to be installed.

$ ./contrib/build-wheel.sh

Running with ETCD

NIXL can use ETCD for metadata exchange between distributed nodes. This is especially useful in containerized or cloud-native environments.

Environment Setup

To use ETCD with NIXL, set the following environment variables:

# Set ETCD endpoints (required) - replace localhost with the hostname of the etcd server
export NIXL_ETCD_ENDPOINTS="http://localhost:2379"

# Set ETCD namespace (optional, defaults to /nixl/agents)
export NIXL_ETCD_NAMESPACE="/nixl/agents"

Running the ETCD Example

NIXL includes an example demonstrating metadata exchange and data transfer using ETCD:

# Start an ETCD server if not already running
# For example:
# docker run -d -p 2379:2379 quay.io/coreos/etcd:v3.5.1

# Set the ETCD env variables as above

# Run the example. The two agents in the example will exchange metadata through ETCD
# and perform data transfers
./<nixl_build_path>/examples/nixl_etcd_example

nixlbench Benchmark

For more comprehensive testing, the nixlbench benchmarking tool supports ETCD for worker coordination:

# Build nixlbench (see benchmark/nixlbench/README.md for details)
cd benchmark/nixlbench
meson setup build && cd build && ninja

# Run benchmark with ETCD
./nixlbench --etcd-endpoints http://localhost:2379 --backend UCX --initiator_seg_type VRAM

Examples

• C++ examples

• Python examples

Third-Party Components

This project will download and install additional third-party open source software projects. Review the license terms of these open source projects before use.