Question

Consider the three main alternate configurations of page tables we have talked about during lecture: hierarchical, hashed, inverted. They are all efforts to address the impacts of very large page tables to physical memory and retrieval latency.

Where do you think does the hierarchical table get stored, where does the hashed page table get stored, and where does the inverted table gets stored? What are the main issues to address? For example, the hierarchical PT consists of a number of inner and outer tables with large indexing. Where would you store all these? How do you think this is done by today’s processors? What about the hashed PT? An element of a hashed page table is not just a single 32-bit word, but much more. How many memory entries does it correspond to? So, where are all these …. extended entries stored? Do we achieve anything by using a hashed table after all?

Answer 1