Implemented modern last-level cache(LLC) with the concept of "Perceptron Learning for Reuse Prediction" that use neural network idea, which is training the predictor by a smaller independent cache with a series of features.

TheGuardian exposes mitigation techniques for cache-based side channel attacks by devising a secure LLC replacement policy. Tests were conducted on ChampSim to make it immune to such cross-core eviction attacks.

Source code for the cache replacement policy published in [Faldu et al., PACT'17] and [Faldu et al., CRC2'17].

Source code for the cache replacement policy published in [Faldu et al., PACT'17] and [Faldu et al., CRC2'17].

A simple piece of software which can infer the size of the Last Level Cache (LLC) of a system via timing memory access.