Lorenzo Ferretti

I am a Ph.D. student in informatics at Università della Svizzera italiana (USI) from July 2016. I am currently working under the supervision of Prof. Laura Pozzi in a research group with Dr. Giovanni Ansaloni, Georgios Zacharopoulos and Ilaria Scarabottolo. I've started working with the research group in January 2016 during my master thesis and in June 2016 I obtained a master degree of Science in Informatics at the Università della Svizzera italiana. Before the master at USI, I obtained a bachelor degree of Science in Informatics in April 2014 at University of Milano-Bicocca. After the bachelor, before starting the master in Switzerland, I have worked for 6 months as research assistant for the IRALab research group at University of Milano-Bicocca.

Research

High Level Synthesis (HLS) frameworks are used to describe hardware designs in high-level languages (C/C++), without burdening developers with the error-prone task of specifying their implementation in detail. The HLS process is usually controlled by user-given directives (e.g., directives to set whether or not to unroll a loop) which influence the resulting implementation area and latency. Nonetheless, the correlation among directives and resulting performance is often difficult to foresee and to quantify, and the high number of available directives leads to an exponential explosion in the number of possible congurations. In addition, sampling the design space involves a time-consuming hardware synthesis, making a brute-force exploration infeasible beyond very simple cases.

For a given design, only few directive settings result in Pareto-optimal solutions (with respect to metrics such as area, run-time and power), while most are dominated. The design space exploration problem aims at identifying Pareto optimal implementations while synthesising only a small portion of the possible congurations from the design space.

Addressing this challenge, my ongoing research effort focuses on the development of methodologies able to smartly navigate the design space to retrieve a close approximation of the real Pareto solutions, synthesising only a small fraction of the possible designs.

Teaching

As Ph.D. student for USI, I have been teaching assistant for the following courses:

Projects

Cluster-Based Design Space Exploration

A cluster-based heuristic that, by only exploring a subset of possible configurations for an High Level Synthesis design, is able to retrieve a close approximation of its Pareto Frontier of non-dominated implementations.
The framework identifies regions of interest in the design space, and iteratively searches for new solutions within such regions, or in their combinations.
More information at this link.

Lattice-Traversing Design Space Exploration

A Lattice-Traversing local search strategy that, by exploring Pareto-solution neighbours, is able to retrieve a close approximation of the Pareto Frontier of a design space.
The framework exploit the property of Pareto-solutions to share high similarity among dominant configurations and iteratively searches for new configuration to imporve the explored Pareto-front.
More information at this link.

Tailoring SVM Inference for Resource-Efficient ECG-Based Epilepsy Monitors

Exploration of multiple optimizations by tailoring Support Vector Machine (SVM) inference engines devoted to the detection of epileptic seizures from ECG-derived features.
The combination of the different optimization strategies resulted in 12.5X and 16X gains in energy and area, respectively, with a negligible loss, 3.2\% in classification performance.
More information at this link.

Lattice-Traversing Design Space Exploration

A novel Design Space Exploration strategy investigating, for the first time, the feasibility of effectively harnessing the knowledge from past synthesis outcomes to guide the optimization of new designs.
The proposed approach, by leveraging prior knowledge from a database of existing design space explorations, dramatically reduces the number of syntheses required by the explorations while retrieving a close approximation of the Pareto frontier.
More information at this link.

Publications

Journals

Master thesis