How to lower computational intensity

[timif2]

What are some ways I can lower the strain/computational intensity of parsing a PDF using Docling? We are trying to deploy a RAG application but the parsing is still quite heavy on our server. So far I have:

• Set pipeline_options.do_ocr = False, set pipeline_options.do_table_structure = False
• Use DoclingParseV2DocumentBackend

Is there anything else we can do?

Separately, I also wanted to ask about the V4 DocumentBackend - would this have any performance/intensity changes compared to say V2? Which backend is puts less strain on CPU?