Speaker: Dr. Yun (Eric) Liang, Assistant Professor, Peking University, China
Title: Enabling Effective Performance Optimization Techniques for Heterogeneous System
Heterogeneous systems that couple CPUs with accelerators such as Graphics Processing Units (GPUs) and Field Programmable Gate Arrays (FPGAs) have become ubiquitous in the computing world due to their tremendous computing power. However, performance tuning for GPUs and FPGAs is not trivial. In this talk, I will first present the performance modeling and optimization for FPGAs. I will introduce an accurate performance model for OpenCL workloads on FPGAs and how we accelerate deep learning applications on FPGAs. For the second half of the talk, I will present on-chip storage and machine learning optimization techniques for GPUs. The proposed techniques leverage on compile-time and run-time techniques to improve the cache performance, register utilization, pipeline utilization and overall performance.
Yun (Eric) Liang is an assistant professor in School of EECS, Peking University, China. He received his PhD in Computer Science from the National University of Singapore and worked as a Research Scientist in UIUC before he joins PKU. His research focuses on energy-efficient heterogeneous computing, computer architecture, compilation techniques, electronic design automation and embedded system design. He has authored over 60 scientific publications in premier international journals and conferences in this domain. His research has been recognized by best paper award at FCCM 2011 and best paper nominations at ICCAD 2017, DAC 2017, ASPDAC 2016, DAC 2012, FPT 2011, CODES+ISSS 2008. Prof Liang serves as Associate Editor for ACM Transactions in Embedded Computing Systems (TECS) and serves in the program committees in the premier conferences in the related domain including (HPCA, PACT, CGO, ICCAD, ICS, CC, DATE, CASES, ASPDAC, ICCD).