Guangwen Yang

Guangwen Yang

Auto-generated profile, based on publicly available information

Guangwen Yang

Department of Computer Science and Technology, Tsinghua University, Beijing, China

Research Focus & Metrics

Academic Career Length

24Years (2003-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
21 June 2016 in Science China Information Sciences
Verified
Journal Article
Closed access
PDF file
HFHaohuan Fu
Haohuan Fu
JLJunfeng Liao
Junfeng Liao
JYJinzhe Yang
Jinzhe Yang
LWLanning Wang
Lanning Wang
ZSZhenya Song
Zhenya Song
GYGuangwen Yang
Guangwen Yang

13 October 2020 in IEEE Transactions on Parallel and Distributed Systems
Verified
Journal Article
Closed access
MLMingzhen Li
Mingzhen Li
YLYi Liu
Yi Liu
XLXiaoyan Liu
Xiaoyan Liu
QSQingxiao Sun
Qingxiao Sun
XYXin You
Xin You
DQDepei Qian
Depei Qian
The difficulty of deploying various deep learning (DL) models on diverse DL hardware has boosted the research and development of DL compilers in the community. Several DL compilers have been proposed from both industry and academia such as Tensorflow XLA and TVM. Similarly, the DL compilers take the DL models described in different DL frameworks as input, and then generate optimized codes for diverse DL hardware as output. However, none of the existing survey has analyzed the unique design architecture of the DL compilers comprehensively. In this paper, we perform a comprehensive survey of existing DL compilers by dissecting the commonly adopted design in details, with emphasis on the DL oriented multi-level IRs, and frontend/backend optimizations. We present detailed analysis on the design of multi-level IRs and illustrate the commonly adopted optimization techniques. Finally, several insights are highlighted as the potential research directions of DL compiler. This is the first survey paper focusing on the design architecture of DL compilers, which we hope can pave the road for future research towards DL compiler.

25 April 2024 in Neural Networks
Verified
Journal Article
Closed access
YCYushu Chen
Yushu Chen
SLShengzhuo Liu
Shengzhuo Liu
JYJinzhe Yang
Jinzhe Yang
HJHao Jing
Hao Jing
WZWenlai Zhao
Wenlai Zhao
GYGuangwen Yang
Guangwen Yang

30 November 2018 in Nature Methods
Verified
Journal Article
Closed access
PDF file
MHMingxu Hu
Mingxu Hu
HYHongkun Yu
Hongkun Yu
KGKai Gu
Kai Gu
ZWZhao Wang
Zhao Wang
HRHuabin Ruan
Huabin Ruan
XLXueming Li
Xueming Li

1 January 2017 in AIP Conference Proceedings
Verified
Conference Paper
Closed access
YXYifan Xiao
Yifan Xiao
GYGuangwen Yang
Guangwen Yang
We propose a fast method to automatically pick protein particles in cryo-EM micrographs, which is now completed manually in practice. Our method is based on Fast R-CNN, with sliding window as the regions proposal solution. To reduce the false positive detections, we set a single class for the major contaminant ice, and pick out all the ice particles in the whole datasets. Tests on the recently-published cryo-EM data of three proteins have demonstrated that our approach can automatically accomplish the human-level particle picking task, and we successfully reduce the test time from 1.5 minutes of previous deep learning method to 2 seconds without any recall or precision losses. Our program is available under the MIT License at https://github.com/xiao1fan/FastParticlePicker.

1 September 2013
Verified
Conference Paper
Closed access
LGLin Gan
Lin Gan
HFHaohuan Fu
Haohuan Fu
WLWayne Luk
Wayne Luk
CYChao Yang
Chao Yang
WXWei Xue
Wei Xue
GYGuangwen Yang
Guangwen Yang
One of the most essential and challenging components in a climate system model is the atmospheric model. To solve the multi-physical atmospheric equations, developers have to face extremely complex stencil kernels. In this paper, we propose a hybrid CPU-FPGA algorithm that applies single and multiple FPGAs to compute the upwind stencil for the global shallow water equations. Through mixed-precision arithmetic, we manage to build a fully pipelined upwind stencil design on a single FPGA, which can perform 428 floating-point and 235 fixed-point operations per cycle. The CPU-FPGA algorithm using one Virtex-6 FPGA provides 100 times speedup over a 6-core CPU and 4 times speedup over a hybrid node with 12 CPU cores and a Fermi GPU card. The algorithm using four FPGAs provides 330 times speedup over a 6-core CPU; it is also 14 times faster and 9 times more power efficient than the hybrid CPU-GPU node.