Mo Xu

Room 4-301, Rohm Building
Dept. of Electronic Engineering, Tsinghua University
Beijing, 100084, China
Email: xumo08@mails.tsinghua.edu.cn
Mobile: (86)13683637232

I am now a senior undergraduate in Electronic Engineering at Tsinghua University and a research assistant in the Research Institute of Circuits and Systems. My advisor is Dr. Yu Wang. I also take an internship at Hardware Computing Group (HCG), Microsoft Research Asia (MSRA). My mentor is Dr. Ningyi Xu. My research is related to high performance computing on GPUs and multicore CPUs. My CV is available.

Publication

[ICCAD2011] Yu Wang, Mo Xu, Ling Ren, Xiaorui Zhang, Yong He, Ningyi Xu, Huazhong Yang, A Heterogeneous Accelerator Platform for Multi-Subject Voxel-Based Brain Network Analysis, Proceedings of the 2011 IEEE/ACM International Conference on Computer-Aided Design, San Jose, California, Nov. 2011. (Paper Slides)
[IPDPSW2012] Mo Xu, Xiaorui Zhang, Yu Wang, Ziyu Wen, Yi Xu, Ling Ren, Gaolang Gong, Ningyi Xu, Huazhong Yang, Probabilistic Brain Fiber Tractography on GPUs, 26th IEEE International Parallel & Distributed Processing Symposium Workshop on HiCOMB, Shanghai, May. 2012.

Projects

2011.7 - 2012.3 Hardware Acceleration for Ranking Algorithm. In data mining and machine learning applications, weighted histogram calculation often serves as a key component in the processing of their massive data sets. However, the atomic operation, which is introduced to resolve the collisions in GPU-based parallel histogramming with large number of bins, brings the overhead of instruction serialization and limits the performance and performance predictability. Our technical report "Efficient Weighted Histogramming on GPUs with CUDA" describes a hybrid method that dynamically chooses the best solution from the improved traditional method and our new method, and shows the achieved substantial performance gain.
2010.10 - 2011.12 Heterogeneous Brain Network Analysis Accelerator. The research on understanding the human brain has attracted more and more attention. A promising method is to model the brain as a network based on modern imaging technologies and then to apply graph-theoretical algorithms for analysis. The ICCAD'11 paper sumerizes our heterogeneous accelerator toolkit.
2011.7 - 2012.2 GPU-based Brain Fiber Tractography. Diffusion Tensor Magnetic Resonance Imaging (DT-MRI) is an emerging technique that explores the structural connectivity of the human brain.The probabilistic fiber tractography based on DT-MRI data behaves more robustly than deterministic approaches in the presence of fiber crossings, but requires prohibitive computational time. The IPDPSW2012 paper "Probabilistic Brain Fiber Tractography on GPUs" discusses a new task segmenting strategy on GPU for the heavy-tail distributed load, which favors the SIMD architecture while maintaining low CPU-GPU communication overhead.