Go Installation

Use this page when you need the NVIDIA cuVS Go bindings. The Go bindings use CGO and require the native cuVS C and C++ libraries.

All NVIDIA cuVS routine implementations live in the C++ core. The Go bindings call into the native C and C++ libraries, so install both libcuvs_c and libcuvs.

Install Pre-Compiled Dependencies

Install the native libraries first:

$
# CUDA 13
$
conda install -c rapidsai -c conda-forge libcuvs cuda-version=13.2
$
$
# CUDA 12
$
conda install -c rapidsai -c conda-forge libcuvs cuda-version=12.9

When building examples or working on the Go bindings from source, use the centralized build environment guidance. A Go-specific environment YAML file is also available in conda/environments when you want a narrower development environment.

Configure CGO so Go can find the native headers and libraries:

$
export CUDA_HOME="/usr/local/cuda"
$
export CGO_CFLAGS="-I${CONDA_PREFIX}/include -I${CUDA_HOME}/include"
$
export CGO_LDFLAGS="-L${CONDA_PREFIX}/lib -lcuvs -lcuvs_c"
$
export LD_LIBRARY_PATH="$CONDA_PREFIX/lib:$LD_LIBRARY_PATH"
$
export CC=clang

Add the Go module to your project:

$
go get github.com/rapidsai/cuvs/go@v26.06.00

Then build your project with the usual Go tooling:

$
go build ./...

Build From Source

Before building from source, review the shared C++ source-build prerequisites, including the recommended conda environment setup for build dependencies.

Build the native libraries first if needed:

$
./build.sh libcuvs

Then build the Go bindings:

$
./build.sh go

The Go bindings require CGO, so builds will fail if the native libraries are not installed or if the CGO include and library paths are not configured correctly.