Is there an option to store values in smaller data types instead of double?

[HariniMurugan11]

I’m working with RRDtool for storing time-series data. I noticed that all data points are stored as double values (8 bytes).

For large datasets, this consumes more memory compared to other lightweight TSDBs (e.g., NanoTS, InfluxDB, etc.), especially when we don’t need high precision.

My question is:

• Is there any existing option in RRDtool to store values in a smaller data type (e.g., float or integer)?
• Or is using double fixed by design?
• If not configurable, do you recommend any workaround to reduce memory/disk usage?