Scientific workloads often involve very large datasets. It might be high resolution data captured from various sensors, or it might be more “normal” data but in huge quantities. Assuming the data itself is high quality, larger datasets mean more accurate conclusions.