M04 Modern High-Level Synthesis for Complex Data Science Applications

Start
Monday, 21 March 2022 13:15
End
Monday, 21 March 2022 17:15
Organizer
Serena Curzel, Pacific Northwest National Laboratory, US and Politecnico di Milano, Italy
Organizer
Nicolas Bohm Agostini, Pacific Northwest National Laboratory and Northeastern University, United States
Organizer
Michele Fiorito, Politecnico di Milano, Italy
Organizer
Marco Minutoli, Pacific Northwest National Laboratory, United States
Organizer
Vito Giovanni Castellana, Pacific Northwest National Laboratory, United States
Organizer
Fabrizio Ferrandi, Politecnico di Milano, Italy
Organizer
Antonino Tumeo, Pacific Northwest National Laboratory, United States

Data Science applications (machine learning, graph analytics) today are the main drivers for designing domain-specific accelerators, both for reconfigurable devices such as Field Programmable Gate Arrays (FPGAs) and Application-Specific Integrated Circuits (ASICs). As data analysis and machine learning methods keep evolving, we are experiencing a renewed interest in high-level synthesis (HLS) and automated accelerator generation to reduce development effort and allow quick transition from the algorithmic formulation to hardware implementation. This tutorial will discuss the use of modern HLS techniques to generate domain-specific accelerators, explicitly focusing on accelerators for data science, highlighting key methodologies, trends, advantages, benefits, and gaps that still need to be closed. The tutorial will provide a direct hands-on experience with Bambu, one of the most advanced open-source HLS tools currently available, and SODA-OPT, an open-source frontend tool for HLS developed in MLIR. Bambu supports many logic synthesis and simulation tools by integrating various compiler frontends, generating accelerators targeting a variety of FPGA devices and ASIC flows, and introducing new methodologies for parallel accelerators (dataflow and multithreaded designs). SODA-OPT performs hardware/software partitioning of specifications derived from popular high-level data science and machine learning Python frameworks used in high-level data-driven applications. Additionally, it provides domain-specific optimizations to improve the high-level synthesis process of the identified hardware components. Integrating SODA-OPT with Bambu allows the generation of highly efficient accelerators for complex graph analysis and machine learning algorithms.

Intro-1 Agile Hardware Design for Complex Data Science Applications: Opportunities and Challenges.

Session Start
Mon, 13:15
Session End
Mon, 13:45
Speaker
Antonino Tumeo, Pacific Northwest National Laboratory, United States

Intro-2 Bambu: an Open-Source Research Framework for the High-Level Synthesis of Complex Applications.

Session Start
Mon, 13:45
Session End
Mon, 14:15
Speaker
Fabrizio Ferrandi, Politecnico di Milano, Italy

Hands-on-1 Productive High-Level Synthesis with Bambu

Session Start
Mon, 14:15
Session End
Mon, 15:00
Speaker
Serena Curzel, Pacific Northwest National Laboratory, US and Politecnico di Milano, Italy

Hands-on-2 Compiler Based Optimizations, Tuning and Customization of Generated Accelerators

Session Start
Mon, 15:15
Session End
Mon, 16:00
Speaker
Michele Fiorito, Politecnico di Milano, Italy

Hands-on-3 SODA-OPT: Enabling System-Level Design in MLIR for High-Level Synthesis and Beyond

Session Start
Mon, 16:00
Session End
Mon, 16:45
Speaker
Nicolas Bohm Agostini, Pacific Northwest National Laboratory and Northeastern University, United States

Tech-1 Svelto: High-Level Synthesis of Multi-Threaded Accelerators for Graph Analytics

Session Start
Mon, 16:45
Session End
Mon, 17:15
Speaker
Marco Minutoli, Pacific Northwest National Laboratory, United States
Speaker
Vito Giovanni Castellana, Pacific Northwest National Laboratory, United States