> I would allow 1k per
> triple (depending on literal sizes)
Since adding the intern cache to parsing, the memory footprint has dropped.
We use 500 bytes/triple and that has a generous over-estimate for
temporary processing space that is over and above storage.
Some measuring of storage requirement [1] (not inferencing):
RxNORM: 180 bytes/triple.
SNOMED CT: 212
chembl_21.0_unichem: 280
Obviously many long literals will throw that rough estimate.
Andy
[1] Measured with:
https://github.com/afs/jena-workspace/blob/master/src/main/java/tools/RunMemTimeSpace.java
See FactoryRDFCaching
