You can profile memory usage of any Daft query using Memray: https://github.com/bloomberg/memray (only works on Linux and Mac)

We have a Pytest fixture that allow you to profile memory usage in a benchmark test: https://github.com/Eventual-Inc/Daft/blob/d632c7a58cd6fe96148f345e24b9063ec3795451/tests/benchmarks/conftest.py#L44-L58

Example output for running TPCH Q6: