Integrating design space exploration in modern compilation toolchains
keywords ARTIFICIAL INTELLIGENCE, C, COMPILERS, CONVOLUTIONAL NEURAL NETWORKS, DEEP LEARNING, DEEP NEURAL NETWORKS, DESIGN SPACE EXPLORATION, EMBEDDED SYSTEMS, ENERGY EFFICIENCY, FIRMWARE, HARDWARE ACCELERATORS, LOW POWER, MICROCONTROLLERS, SOFTWARE, SOFTWARE ACCELERATION
Reference persons DANIELE JAHIER PAGLIARI
External reference persons Alessio Burrello (Politecnico di Torino)
Marian Verhelst (KU Leuven)
Thesis type EXPERIMENTAL, SOFTWARE DEVELOPMENT
Description Modern compilers have shown great advancements in optimizing code deployment, minimizing data transfers and optimize the energy consumption of the target platform.
In the same direction, new design space exploration (DSE) tools have been created to explore the vast space of deep neural network execution and hardware accelerators for their execution. In particular, in modern solutions, there is a missing link between what DSE tools produce and what is deployable on hardware.
In this thesis, the candidate will explore two tools, ZigZag, a powerful DSE tools used for neural network loop ordering and tiling and hardware accelerators-NN schedule co-design and HTVM, a tool for compilation which exploit the open-source TVM tool, integrated with a custom plug-in to produce optimize code for heterogeneous platforms.
The goal of the candidate would be to create an interface between the two tools and allow the deployment of ZigZag produced solution, to enhance the performance of NN execution on edge devices.
Required skills Required skills include C and python programming. Further, a basic knowledge of computer architectures and embedded systems is necessary. Desired (but not required) skills include some familiarity with basic machine/deep learning concepts and corresponding models.
Notes Thesis in collaboration with Prof. Marian Verhelst’s research group at KU Leuven. The thesis can be carried out either in Torino or in Leuven, Belgium.
Deadline 13/06/2022 PROPONI LA TUA CANDIDATURA