I think you can assume it works - but I would also suggest if you're hitting that sort of data size you may want to explore Spark or Dask as an execution engine