Benchmarks and evaluation code for the IPDPS26 submission on Academy.
The bench/ package contains all of the microbenchmarks for the paper.
python -m venv venv
. venv/bin/activate
pip install -e .This installs the requirements for notebooks/ as well, but the analysis specific requirements are also defined in notebooks/requirements.txt.
Each benchmark is an executable module in the bench package. E.g.,
python -m bench.action_latency --helpAll results used in the paper are provided in data/, and all figures have an associated Jupyter Notebook in notebooks/.