Operating Systems Laboratory

/ Minyi Guo / Assistant Professor

In the Operating Systems Laboratory, the main research programs are parallel and distributed operating systems, parallelizing compilers, and dynamic compilers.

In education, the laboratory undertakes the undergraduate courses ``Operating Systems" and ``Computer Architecture (Seminar)", and graduate courses ``Parallel and Distributed Processing (I), (II)". With respect to Operating Systems, we have installed an operating system simulation environment Nachos, and set of assignments that we developed for the undergraduate operating systems course.

For the graduate courses, we constructed the parallel environments on CM-5 and workstation cluster which can use parallel library MPI, and taught some topics about parallel architecture, benchmark, message-passing programming, multithread programming, and automatic parallelization, etc.

In research, the laboratory focuses on some of the themes as

1. Optimization for Irregular Applications in Parallelizing Compilers. This research develops techniques and methods to help improve the quality of automatic parallel programs. In detail, we complete the following sub-project:
   • Data dependence analysis and loop transformation for irregular applications;
   • Communication set generation for irregular applications;
   • Data distribution for irregular applications.

2. High performance computing environment construction using JAVA. The goal of this theme is to construct a source-to-source restructuring Java compiler. With sequential Java source programs as input the system generates parallel Java source programs for the consequent local javac compiler. The compiler generates Java RMI interfaces for clusters of workstations or parallel machines based on distributed memory system to implement remote method invocation. Target programs implement parallelism with Java threads.

3. Visual parallel programming tools. We are developing a visual tool for representing the behaviors of data dependences, loop transformation, communication, data distribution, and dynamic debug.

4. Design of a virtual machine that integrates the compiler, OS, and processor techniques. The researches of JAVA dynamic compiling, OS kernel and the support mechanism for languages and environments, as well as command set and processor architecture being suitable for object-oriented programming languages are being developed.

Refereed Journal Papers

1. M. Guo, I. Nakata, and Y. Yamashita., Contention-free communication scheduling for array redistribution. Parallel Computing, vol. 26, No. 10, pp. 1325--1343, 2000.

2. M. Guo and I. Nakata., A framework for efficient array redistribution on distributed memory multicomputers. The Journal of Supercomputing, 2001, accepted.

3. M. Guo., Denotational semantic descriptions of data-parallel languages. Parallel Processing Letters, 2001, accepted.

1. M. Guo., A denotational semantic model of an HPF-like language. The First International Conference on Parallel and Distributed Computing, Applications and Technologies, pp. 1-8, Hong Kong University, IEEE Computer Society, Hong Kong, May 2000.

2. M. Guo and I. Nakata., Communication scheduling with considering message size in data redistribution algorithms. The 2000 International Conference on Parallel and Distributed Processing, Techniques and Applications, editor: Hamid Arabnia, PDPTA'2000, IEEE Computer Society, Las Vegas, USA., June 2000.

1. Minyi Guo. Reviewer of Parallel Computing (Elsevier), 2000.

2. Minyi Guo. Reviewer of IEICE Trans. Information and Systems, 2000.

3. Minyi Guo. Program committee member of PDPTA'2000, 2000.