Installation¶

Prerequisites¶

Python 3.10+
PostgreSQL (for Prefect Orion database)

Setting up PostgreSQL¶

Install PostgreSQL:

macOS (Homebrew):

brew install postgresql
brew services start postgresql

Linux (Debian/Ubuntu):

sudo apt-get update && sudo apt-get install -y postgresql postgresql-contrib
sudo service postgresql start

Windows: Download and run the PostgreSQL installer and follow the prompts.

Create a PostgreSQL User and Database:

psql postgres

Inside the psql prompt:

CREATE USER prefect_user WITH PASSWORD 'pass';
CREATE DATABASE prefect_db OWNER prefect_user;
GRANT ALL PRIVILEGES ON DATABASE prefect_db TO prefect_user;

Exit with \q.

Installing nxbench¶

From PyPi:

pip install nxbench

From a local clone:

git clone https://github.com/dpys/nxbench.git
cd nxbench
make install

Setting up Prefect Orion¶

In a terminal window:

Export environment variables pointing to your PostgreSQL database:

export PREFECT_API_DATABASE_CONNECTION_URL="postgresql+asyncpg://prefect_user:pass@localhost:5432/prefect_db"
export PREFECT_ORION_DATABASE_CONNECTION_POOL_SIZE="5"
export PREFECT_ORION_DATABASE_CONNECTION_MAX_OVERFLOW="10"
export PREFECT_ORION_API_ENABLE_TASK_RUN_DATA_PERSISTENCE="false"
export PREFECT_API_URL="http://127.0.0.1:4200/api"

Start the Orion server:
```
prefect orion start
```
By default it will run on http://127.0.0.1:4200.

Running Benchmarks¶

In a new terminal window:

nxbench --config 'nxbench/configs/example.yaml' benchmark run

After the run completes, you can export results:

nxbench --config 'nxbench/configs/example.yaml' benchmark export 'results/<run_file>.json' --output-format csv --output-file 'results/results.csv'

To visualize:

nxbench viz serve

Installation (Docker Setup)¶

# CPU-only
docker-compose -f docker/docker-compose.cpu.yaml build

# With GPU
docker-compose -f docker/docker-compose.gpu.yaml build

Development Setup¶

Install development dependencies (testing and documentation):

pip install -e .[test,doc]

Run tests:

make test