Experience

 
 
 
 
 
June 2018 – August 2018
California

Research Intern

Facebook Inc

Neural Network Optimization on Mobiles
 
 
 
 
 
April 2017 – July 2017
Beijing

Research Intern

Nvidia Inc

Neural Network Quantization
 
 
 
 
 
October 2013 – February 2014
Beijing

SDE Intern

Baidu Inc

Hadoop Workflow Optimization

Projects

*

Our tool provides a profile view and a trace view for GPU-accelerated applications. The profile view identifies where GPU APIs are invoked in CPU calling context, approximates calling context for GPU execution, and analyzes instruction mix for GPU kernels. The tool traces CPU and GPU activities for a large number of processes and threads with minimal overhead.

A fast, memory efficient, and light-weight implementation for gSpan algorithm in data mining. gBolt is up to 100x faster than the original implementation with multi-threading on a single machine. gBolt also reduces more than 200 folds memory usage, running efficiently on personal computers.

Recent Publications

Quickly discover relevant content by filtering publications.

(2019). A tool for performance analysis of GPU-accelerated applications. Proceedings of the 2019 IEEE/ACM International Symposium on Code Generation and Optimization (CGO’19).

Source Document

(2018). Quadboost: A Scalable Concurrent Quadtree. IEEE Transactions on Parallel and Distributed Systems (TPDS’18).

Source Document

(2017). A performance analysis framework for exploiting GPU microarchitectural capability. Proceedings of the International Conference on Supercomputing (ICS’17).

Source Document

(2017). Understanding the GPU microarchitecture to achieve bare-metal performance tuning. ACM SIGPLAN Notices (PPoPP’17).

Source Document

(2015). BF-MapReduce: A Bloom Filter Based Efficient Lightweight Search. 2015 IEEE Conference on Collaboration and Internet Computing (CIC’15).

Source Document

See all publications

Recent & Upcoming Talks

Presented our GPU performance tool
Jul 29, 2019 12:00 AM Lake Tahoe, California
Slides

Presented the prototype of our GPU performance tool
Feb 20, 2019 12:00 AM Washington, D.C.
Poster Slides

Presented our ICS’17 work.
Jun 23, 2017 12:00 AM Northwestern University, Chicago
Slides

Discussed how state-of-the-art deep learning libraries optimize computations by utilizing architectural features.
Apr 1, 2017 12:00 AM Institute of Computing Technology, Chinese Academy of Sciences
Slides

Introduced various kinds of convolution methods and analyzed their complexities, memory consumptions, and data access patterns.
Jun 1, 2016 12:00 AM Institute of Computing Technology, Chinese Academy of Sciences
Slides

News

Topics

CPU Concurrent Algorithm Data Mining Data Structure Deep Learning GPU Parallel Computing Performance Tool