Bin Ren Department of Computer Science & Engineering The Ohio State University 778 Dreese Lab Email: ren [at] cse.ohio-state.edu

• Currently, I'm working on:

 i. Irregular I/O optimization using prefetching method with Professor Agrawal;

ii. Irregular data structure optimization by data parallelism and operation parallelism with Dr. Mytkowicz in Microsoft Research.

• Compiling Dynamic Data Structures in Python to Adapt HPC Applications
  
  Python is a well-known programming language for its high productivity and short study curve. Its performance is the main obstacle for widely usage in HPC community. In this work, my main objective is to combine Python with low level HPC infrastructures to achieve both easy programming and high performance. Our contribution includes:
  
  i.   Build a translation framework to translate the dynamic data structures in Python to low level dense data structure that can be used by low level multi-core/many-core HPC environments;
  ii.  Design a novel Inter-procedural Partial Redundancy Elimination algorithm to reduce the overhead during the data structure transformation;
  iii. Implement the framework and evaluate it by K-Means, PCA, DGEMM and Tensor Multiplication applications in FREERIDE multi-core and CUDA environments.

  Related Publication:

  Bin Ren, Gagan Agrawal, Compiling Dynamic Data Structure in Python to Enable the Use of Multi-core and Many-core Libraries, the 20th International Conference on Parallel Architectures and Compilation Techniques (PACT), October, 2011 (Accept Rate: 16%).

• Compiling Chapel down to FREERIDE Middle-Ware
  
  Chapel is a high-level programming language developed by Cray Inc. and FREERIDE is a Map-Reduce like High Performance Computing (HPC) middle-ware for data-mining and data-intensive applications. To utilize the high productivity of Chapel and the high performance of FREERIDE, a translation mechanism based on compilation technology is established. In this project, our contribution includes:
  
  i.   Present a case study for the possible use of a new HPC language, Chapel, for data-intensive computations;
  ii.  Develop a linearization-mapping algorithm to translate the high level data structure in Chapel to the low level dense data structure that can be used for FREERIDE;
  iii. Build a compilation framework and evaluate it by two data mining algorithms: K-Means and PCA.

Related Publication:

Bin Ren, Gagan Agrawal, Brad Chamberlain, Steve Deitz, Translating Chapel to Use FREERIDE: A Case Study in Using an HPC language for Data-intensive Computing, the 16th International Workshop on High-Level Parallel Programming Models and Supportive Environments (HIPS) held in conjunction with IPDPS, May, 2011, (LCPC2010 Poster).[pdf]

• Spring 2011, CSE756 Compiler Design and Implementation, grader;
• Winter 2011, CSE721 Introduction to Parallel Computing, grader;
• Autumn 2010, CSE621 High Performance Computing, grader;
• Summer 2010, CSE459.23 Programming in Java, CSE459.22 Programming in C++, instructor; 
• Spring 2010, CSE459.22 Programming in C++, instructor;
• Winter 2010, CSE459.23 Programming in Java , instructor;
• Autumn 2009, CSE621 High Performance Computing, grader;