[2019-05-17] Ms. Wei-Fen Lin, Computer Architect and VP of Engineering, Skymizer Taiwan Inc.,

Poster:Ms Zhōu Chéng-Mǎn (Ms Cheng-Man Chou)Post date:2019-04-29
Title: Reshaping AI Chip Design with the Open Neural Network Compiler (ONNC)
Date: 2019-05-17 2:20pm-3:30pm
Location: R103, CSIE
Speaker: Ms. Wei-Fen Lin, Computer Architect and VP of Engineering, Skymizer Taiwan Inc.
Hosted by: Prof. Chia-Lin Yang


The frenzy of global race in artificial intelligence (AI) research and development has been underway in the past two years. Governments, enterprises and research community pour tremendous amount of resources into creating the best architectures and systems to process huge volumes of data as patterns rather than individual bits for AI models, effectively raising the abstraction level for computing and increasing the density of the software. Data movement. computational and programming efficiency are the major driving forces for the state-of-the-art AI chip design. It is a segment where software/hardware co-design is a necessity , but much of the software is far behind the hardware. This talk introduces Open Neural Network Compiler (ONNC), a retargetable compilation framework designed specifically for proprietary deep learning accelerators. ONNC is the first open source compiler project designed from ground up to support Open Neural Network eXchange (ONNX) models. ONNX enables interchangeability among neural network models designed in different frameworks and has become a pervasive format supported by many high-tech titans and a community of researchers. This talk will cover the overview of ONNC, how ONNC helps AI chip design in a product design cycle, and how its integrated system-level software platform reshapes the AI chip design flow. Support for Nvidia Deep Learning Accelerator (NVDLA) and computing in memory (CIM) will also be discussed for illustration.
Wei-Fen Lin, Computer Architect and VP of Engineering, Skymizer Taiwan Inc. Wei-Fen Lin has 15+ years of work experience in the fields of computer architecture and SoC design. Her research interests include high-performance microprocessor design, embedded system and software/hardware co-design. Her most recent work focus is on system design, performance analysis and optimization for AI chip design. She is also the founder of Play Lab, an non-profit private organization for promoting technology and STEAM education to K12 kids and training young engineers.
Last modification time:2019-05-17 AM 11:43

cron web_use_log