Installation

Choose the installation method that best fits your environment.

Pip (Recommended)

The pip package is the easiest way to get CLIO Core. It ships a portable, self-contained build with everything statically linked, so no system dependencies are required beyond a standard C/C++ runtime (glibc) and Python 3.10+.

pip install iowarp-core

Verify the installation:

import iowarp_core
print(iowarp_core.get_version())

Start the Runtime

The package installs the clio_run CLI on your PATH:

clio_run start

A default ~/.clio/clio.yaml is seeded on first import so the runtime works out of the box. See the Quick Start for the next steps.

What's Included

The pip wheel ships the portable feature set:

• Python API — import iowarp_core and import clio_cee for the Context Exploration Engine
• CLI — clio_run command for runtime management
• Core engines — CLIO Runtime, CTE, CAE, CEE
• Shared libraries — every CLIO Runtime library bundled in the wheel; no system installs needed

When You'll Need a Source Build

Newer extensions and advanced/accelerated features are not in the portable wheel. Switch to one of the source-build methods below if you need:

• NVIDIA GPU (CUDA) or AMD GPU (ROCm) acceleration
• MPI for distributed multi-node deployment
• HDF5 scientific data ingestion
• ADIOS2 adapter for streaming analytics
• Compression backends (LibPressio, Blosc, etc.)
• FUSE adapter
• Custom ChiMods you intend to compile against the C++ headers
• Sanitizer / debug builds for development

Conda

The Conda installer builds CLIO Core from source with all optional components enabled. It is the fastest path to a full-featured build that includes GPU, MPI, HDF5, and adapter support.

git clone --recurse-submodules https://github.com/iowarp/clio-core.git
cd clio-core
./install.sh

Activate the environment in every new terminal:

conda activate iowarp

Build Variants

Pass a variant name to enable additional features:

Variant	Command	What it enables
Release (default)	./install.sh	CPU-only, all engines
Debug	./install.sh debug	Debug symbols, sanitizers
CUDA	./install.sh cuda	NVIDIA GPU acceleration
ROCm	./install.sh rocm	AMD GPU acceleration
MPI	./install.sh mpi	Distributed multi-node
Full	./install.sh full	CUDA + MPI + everything
FUSE	./install.sh release-fuse	FUSE adapter for POSIX

Verify the installation

conda activate iowarp
clio_run --help

Docker

Pull and run the CLIO Core Docker image:

docker pull iowarp/deploy-cpu:latest

Run the container:

docker run -d -p 9413:9413 --memory=8g --name iowarp iowarp/deploy-cpu:latest clio_run start

Using Docker Compose

Create a docker-compose.yml:

services:
  iowarp:
    image: iowarp/deploy-cpu:latest
    container_name: iowarp
    hostname: iowarp
    volumes:
      - ./clio.yaml:/home/iowarp/.clio/clio.yaml:ro
    ports:
      - "9413:9413"
    mem_limit: 8g
    command: ["clio_run", "start"]
    restart: unless-stopped

Start the service:

docker compose up -d

info

CLIO Core uses memfd_create() for shared memory, so no special /dev/shm configuration is needed. Only mem_limit matters for resource control. See the Configuration Reference for details.

Spack

1. Install Spack (v0.22.3+ recommended):

git clone https://github.com/spack/spack.git -b v0.22.3
cd spack
. share/spack/setup-env.sh
echo ". ${PWD}/share/spack/setup-env.sh" >> ~/.bashrc

2. Add the IOWarp Spack repository (included in this repo):

git clone --recurse-submodules https://github.com/iowarp/clio-core.git
cd clio-core
spack repo add installers/spack

3. Install CLIO Core:

spack install iowarp

Next Steps

• Quick Start Tutorial — Start the runtime and run your first example
• Configuration Reference — Customize your deployment
• CLIO Kit — Explore MCP servers for AI agents