2019 International Conference On Computer Aided Design

The Premier Conference Devoted to Technical Innovations in Electronic Design Automation

November 4-7, 2019The Westin Westminster Westminster, CO

v class="event-details"> MP Associates, Inc.

TUESDAY November 06, 10:30am - 12:00pm | Riveria
EVENT TYPE: REGULAR SESSION

SESSION 5C
High-Performance Deep Learning Accelerators on FPGAs
Moderator:
Yingyan Lin - Rice Univ.
For both edge-devices and cloud servers, FPGA accelerators for deep neural networks have delivered favorable reconfigurability and performance. However, long hardware design time, prior homogeneous designs, and irregularities in deep learning algorithms have limited the achievable throughput and latency. To address these, the papers in this session present automated RTL generation with fine-grained pipelining, hardware-optimized algorithm adaptation, and integration of heterogeneous accelerators.

5C.1*DNNBuilder: An Automated Tool for Building High-Performance DNN Hardware Accelerators for FPGAs
 Speaker: Xiaofan Zhang - Univ. of Illinois at Urbana-Champaign
 Authors: Xiaofan Zhang - Univ. of Illinois at Urbana-Champaign
Junsong Wang - IBM Research - China
Chao Zhu - IBM Research - China
Yonghua Lin - IBM Research - China
Jinjun Xiong - IBM T.J. Watson Research Center
Wen-Mei Hwu - Univ. of Illinois at Urbana-Champaign
Deming Chen - Univ. of Illinois at Urbana-Champaign
5C.2Algorithm-Hardware Co-Design of Single Shot Detector for Fast Object Detection on FPGAs
 Speaker: Yufei Ma - Arizona State Univ.
 Authors: Yufei Ma - Arizona State Univ.
Tu Zheng - Fuzhou Univ.
Yu Cao - Arizona State Univ.
Sarma Vrudhula - Arizona State Univ.
Jae-Sun Seo - Arizona State Univ.
5C.3TGPA: Tile-Grained Pipeline Architecture for Low Latency CNN Inference
 Speaker: Xuechao Wei - Peking Univ.
 Authors: Xuechao Wei - Peking Univ.
Yun (Eric) Liang - Peking Univ.
Xiuhong Li - Peking University, Beijing
Cody Hao Yu - Univ. of California, Los Angeles
Peng Zhang - Falcon Computing Solutions, Inc.
Jason Cong - Univ. of California, Los Angeles


* Indicates Best Paper Candidate