This repo is my undergrad research project report, on designing efficient data structures by leveraging memory level parallelism, a CPU hardware feature available on most of the modern CPUs manufactured after 2000. pdf here

In this report, we demostrated that by leveraging memory level parallelism, one can design data structures that is up to 4x faster than the current state-of-the-art.

We present two novel "MLP-grounded" data structures:

• A lock-free skiplist iterator
• An ordered index

Those data structures support identical APIs as their traditional counterparts, specifically, std::set::iterator and std::set.

The implementation of our ordered index can be found here.