Page 74 - profile-ok

研究人員 ｜ Research Faculty

● Research Fellow, Academia Sinica, Taiwan (1998-present)
李丕榮 PeiZong Lee ● Associate Research Fellow, Academia Sinica, Taiwan (1989-1998)
● Research Engineer, Institute of Information Industry, Taiwan (1984-
1986)
● Ph.D. , Computer Science , Courant Institute of Mathematical Sci-
研究員 Research Fellow ences, New York University, United States (1989)
Ph.D., Computer Science, New York University ● M.S. , Computer and Decision Sciences , National Tsing Hua Univer-
sity, Taiwan (1984)
Tel: +886-2-2788-3799 ext. 1812 ● B.S. , Mathematics , National Taiwan University, Taiwan (1979)
Fax: +886-2-2782-4814
Email: leepe@iis.sinica.edu.tw
http://www.iis.sinica.edu.tw/pages/leepe

代表著作 Publications

1. PeiZong and Zvi M. Kedem, “Synthesizing Linear-Array Algorithms ume 35, number 2, pages 61--73, February 2004, (Elsevier Science
from Nested For Loop Algorithms,” IEEE Transactions on Comput- B. V.)
研究簡介 Research Description ers: Special Issue on Parallel and Distributed Algorithms, volume 14. PeiZong Lee and Chih-Hao Chang, “Unstructured Mesh Generation
C--37, number 12, pages 1578--1598, December 1988.
Using Automatic Point Insertion and Local Refinement,” Proc. Na-
理想中，平行處理編譯器應該能將循序式程式自動編譯 My research interests are in compilers for scientific applications, parallel algorithm 2. PeiZong Lee and Zvi M. Kedem, “Mapping Nested Loop Algorithms tional Computer Symposium, pages B-550--B-557, Taipei, Taiwan,
成相對的平行處理電腦的執行碼，但是許多刻骨銘心的 design, computer architectures, and the interplay among architectures, algorithms, into Multidimensional Systolic Arrays,” IEEE Transactions on Paral- December 1999.
lel and Distributed Systems, volume 1, number 1, pages 64--76, Janu-
經驗說明要達到自動產生有效率的執行碼，尚待努力。 and compilers. The design of parallel architectures represents cost-effective trade- ary 1990. 15. PeiZong Lee, Chih-Hao Chang, and Maw-Jyi Chao, “A Parallel Euler
th
這些原因包括：一個有效率的循序式演算法，可能並無 offs. A general purpose machine requires complex interconnection networks, 3. PeiZong Lee and Zvi M. Kedem, “On High-Speed Computing with Solver on Unstructured Mesh,” Proc. ISCA 13 International Confer-
ence on Parallel and Distributed Computing Systems (PDCS-2000),
法直接轉換並編譯成一個相對地有效率的平行程式碼。 multi-level memory hierarchies, powerful control units, and optimized schemes for a Programmable Linear Array,” Journal of Supercomputing, volume pages 171--177, Las Vegas, Nevada, August 2000.
因為必須整體考慮在整個應用程式中由開始至結尾各個 allocation of machine resources. In contrast, to solve a class of problems from an 4, number 3, pages 223--249, September 1990, (Kluwer Academic 16. PeiZong Lee, Chih-Hao Chang, and Jan-Jan Wu, “Parallel Implicit
使用到的演算法其分別的資料分割及分佈，資料傳送， application domain or to share a common mathematical structure, it may be possi- Publishers) Euler Solver on Homogeneous and Heterogeneous Computing En-
及執行排程的整體最佳化。 ble to design a special purpose machine with elementary interconnecting networks 4. PeiZong Lee and Fang-Yu Huang, “Restructured Recursive DCT and vironments,” 15 AIAA Computational Fluid Dynamics Conference,
th
and elementary control units. Such a machine may also be able to solve this class DST Algorithms,” IEEE Transactions on Signal Processing, volume Anaheim, CA., June 2001, AIAA paper 2001-2588
對於規則的應用程式，目前已經能夠自動分析整個應用 of problems faster. 42, number 7, pages 1600--1609, July 1994.
程式的資料流程，並且可以產生有效率的執行碼。但是 5. PeiZong Lee and Fang-Yu Huang, “An Efficient Prime-Factor Al- 17. PeiZong Lee, Jan-Jan Wu, and Chih-Hao Chang, “Partitioning Un-
structured Meshes for Homogeneous and Heterogeneous Parallel
對於不規則的應用程式，由於使用到間接位址，尚未有 In the context of parallel algorithms, several paradigms and techniques, such as gorithm for the Discrete Cosine Transform and Its Hardware Imple- Computing Environments,” Proc. International Conference on Paral-
較有效率的最佳化資料分析技術。目前較可行的方法包 divide and conquer, dynamic programming, domain decomposition, and program mentations,” IEEE Transactions on Signal Processing, volume 42, lel Processing, pages 315--322, Vancouver, British Columbia, August
括由程式語言提供工作元層面的指令集，並且由程式員 transformation, have been studied. By using these paradigms, it is possible to de- number 8, pages 1996--2005, August 1994. 2002.
提供應用問題層面的大區塊領域分割演算法，編譯器再 sign efficient parallel algorithms, at least in theory. Their practical utilization requires 6. PeiZong Lee, “Mapping Nested-Loop Algorithms into Grid-Connect- 18. PeiZong Lee, Chih-Hsueh Yang, and Jeng-Renn Yang, “Repartition-
將領域分割演算法編譯至平行處理器上執行。由於實用 appropriate architectural and software support, techniques for proving the correct- ed Systolic Arrays without Data Collisions in the Data Links,” ISCA ing Unstructured Meshes for the Parallel Solution of Engine Combus-
的平行處理電腦未來的趨勢是收歛至網格狀的大量平行 ness and efficiency of implementation, and suitable programming environments. International Journal of Computers and Their Applications, volume tion,” Supercomputing’2003, Phoenix, AZ, November 2003.
處理器或由網路連接的叢集多處理器電腦工作站，未來 The design of parallel architectures influences the design of parallel algorithms, and 2, number 2, pages 72--85, August 1995, (International Society for 19. PeiZong Lee, Chien-Min Wang, and Jan-Jan Wu. Compiler and
Computers and Their Applications)
平行處理的成功與否將僅決定在是否能整合平行演算法 vice versa. Compilers act as bridges connecting algorithms and architectures. I am Run-time Parallelization Techniques for Scientific Computations on
及編譯器至此兩大類平行處理電腦上。 7. PeiZong Lee, “Parallel Matrix Multiplication Algorithms on Hyper- Distributed Memory Parallel Computers. In book High Performance
interested in studying this interdependence. I have studied various compilation is- cube Multicomputers,” International Journal of High Speed Comput- Computing: Paradigm and Infrastructure, Laurence T. Yang and
在未來的研究方面，我將致力於發展一個數值風洞模擬 sues on regular parallel architectures, focusing on formal techniques, simulations, ing, volume 7, number 3, pages 391--406, September 1995, (Kluwer Minyi Guo, editor, pages 135--181, John Wiley & Sons, Inc., 2006.
Academic Publishers)
全機流場的軟體環境平台以及一個模擬引擎燃燒反應流 and the development of programming environments. The results obtained can be 20. Chih-Hung Hsu and PeiZong Lee. “Study of Flow Caculations at All
th
的軟體環境平台。我希望能暸解平行演算法，編譯器， used both to compile algorithms on such regular parallel architectures and to study 8. PeiZong Lee, “Techniques for Compiling Programs on Distributed Speeds on Unstructured Grids,” the 14 National Computational Flu-
id Dynamics Conference, Nantou, Taiwan, Aug. 16-18, 2007.
Memory Multicomputers,” Parallel Computing, volume 21, number
以及大型不規則科學計算應用問題三者之間的互動關 the importance of various architectural features. Current effort also has extended 12, pages 1895--1923, December 1995, (Elsevier Science B. V.)
係。 this approach to the compilation of large scientific applications for modern general- 21. Chih-Hsueh Yang, PeiZong Lee, Yeh-Ching Chung, “Improving Stat-
purpose regular architectures. 9. PeiZong Lee and Gau-Shin Liu, “An Efficient Algorithm for the 2-D ic Task Scheduling in Heterogeneous and Homogeneous Computing
Discrete Cosine Transform,” Signal Processing, volume 55, number Systems,” Proc. International Conference on Parallel Processing,
I am currently also participating in a scientific computation project, in which some 2, pages 221--239, December 1996, (Elsevier Science B. V.) ICPP paper 2007-45, Xi-An, China, September 10-14, 2007.
colleagues and I have to conquer researches for unstructured mesh generation, 10. PeiZong Lee, “Efficient Algorithms for Data Distribution on Distrib- 22. Chih-Hung Hsu, PeiZong Lee, Juan-Hong Hong, Yu-Ting Deng, and
unstructured mesh partition, Euler equation and Navier-Stokes equation solvers uted Memory Parallel Computers,” IEEE Transactions on Parallel Keh-Chin Chang. “Simulation on Rapid Expansion of Supercritical
th
for a numerical wind tunnel platform, an engine combustion platform for comput- and Distributed Systems, volume 8, number 8, pages 825--839, Au- Carbon Dioxide,” 16 National Computational Fluid Dynamics Con-
ference, Yilan, Taiwan, July30 -- Aug. 1, 2009.
gust 1997.
ing reactive flows, visualization, and the challenge of using MPI on workstation/PC
clusters to accelerate irregular computation. One of my goals is to understand the 11. PeiZong Lee and Zvi M. Kedem, “Automatic Data and Computation
Decomposition on Distributed Memory Parallel Computers,” ACM
techniques for compiling irregular computation. Transactions on Programming Languages and Systems, volume 24,
number 1, pages 1--50, January 2002.
12. PeiZong Lee and Wen-Yao Chen, “Generating Communication Sets
of Array Assignment Statements for Block-cyclic Distribution on
Distributed Memory Parallel Computers,” Parallel Computing, vol-
ume 28, number 9, pages 1329--1368, September 2002, (Elsevier Sci-
ence B. V.)
13. PeiZong Lee, Chih-Hsueh Yang, and Jeng-Renn Yang, “Fast Algo-
rithms for Computing Self-Avoiding Walks and Mesh Intersections
over Unstructured Meshes,” Advances in Engineering Software, vol-

74 75