Fig. 1 | Scientific Data

Fig. 1

From: M100 ExaData: a data collection campaign on the CINECA’s Marconi100 Tier-0 supercomputer

Fig. 1

Loading times comparison across compression configurations (Parquet). The data is 4 months of gpu0_core_temp (IPMI), from May to August 2022 (included), retrieving the “timestamp”, “value” and “node” columns. Memory usage by the Pandas Dataframe is 6.55 GB in all cases. The PyArrow io_threads and cpu_count were both set to 8 (Intel Xeon Gold 5220 CPU). The data was loaded using the PyArrow Dataset API.

Back to article page