Co-Operative Research on Performance Evaluation in Supercomputing

超级计算性能评估合作研究

• 批准号: 03302035
• 负责人: SHIMASAKI Masaaki
• 金额: $ 9.22万
• 依托单位: KYUSYU UNIVERSITY
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Co-operative Research (A)
• 财政年份: 1991
• 资助国家: 日本
• 起止时间: 1991 至 1993
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-03302035/
• 关键词: supercomputers; benchmarks; mathematical functions; vectorizing compilers; indirect load; store; parallel computing; computer architecture; application programs; Benchmarking; Computers; FORTRAN; Fluid Dynamics; Namcucel Lincas Algchia; Para

It is important and effective to investgate performance evaluation method of supercomputers, from view point of (1)computer architecture, (2)systems programs including operating systems and compilers, (3)basic library routines, (4)application programs. Research results are largely reported in ISS'91(International Symposium on Supercomputing, Kyusyu Univ.1991)and WBPE : (Workshop on Benchmarking and Performance Evaluation in High Performance Computing, Tokyo, 1993). Goto and Oyanagi proposed six benchmarks for performansce evaluation of supercomputer architectures and reported the result of performance of Hitachi S-3800 using their benchmarks. Tsuda et.al.reported performance evaluation of amemory reference, especially vector indirect load/store instructions. Nagai and Hatano reported results of performance evaluation of mathematical functions including analysis of dependency on function arguments. For performance evaluation at the level of application programs, various problems in specific application field were selected : among them are : fluid dynamics, computer graphics, electrical engineering and chemistry.

从(1)计算机体系结构、(2)包括操作系统和编译器在内的系统程序、(3)基本库例程、(4)应用程序的角度来研究超级计算机的性能评估方法是重要而有效的。研究成果主要在 ISS'91（国际超级计算研讨会，九州大学，1991 年）和 WBPE：（高性能计算基准和性能评估研讨会，东京，1993 年）中报告。 Goto和Oyanagi提出了超级计算机架构性能评估的六个基准，并报告了使用他们的基准的Hitachi S-3800的性能结果。 Tsuda 等人报告了内存引用的性能评估，特别是向量间接加载/存储指令。 Nagai 和 Hatano 报告了数学函数的性能评估结果，包括对函数参数依赖性的分析。对于应用程序级别的性能评估，选择了特定应用领域的各种问题：其中包括：流体动力学、计算机图形学、电气工程和化学。

项目成果

M.Shimasaki,Y.Matsusaka and M.Nomura: "Fractals in Quaternions and their Application to Computer Graphics" IFIP Working Conference on Programing Environments for HighーLevel Scientific Problem Solving. 61-70 (1991)

M. Shimasaki、Y. Matsusaka 和 M. Nomura：“四元数中的分形及其在计算机图形学中的应用”IFIP 高级科学问题解决编程环境工作会议 61-70 (1991)。

T.Nagai: "Benchmarking Fortran Intrinsic Functions" Proc.Workshop on Benchmarking and Performance Evaluation in High Performance Evaluation. 26-32 (1993)

T.Nagai：“Benchmarking Fortran Intrinsic Functions”Proc.Workshop on 高性能评估中的基准测试和性能评估。

T.Hironaka,T.Hashimoto,K.Okazaki,K.Murakami and S.Tomita: "A SingleーChip VectorーProcessor Prototype Based on Multithreaded Streaming/FIFO(MSFV)Architecture" Proceedings of the International Symposium on Supercomputing. 77-86 (1991)

T. Hironaka、T. Hashimoto、K. Okazaki、K. Murakami 和 S. Tomita：“基于多线程流/FIFO (MSFV) 架构的单芯片矢量处理器原型”超级计算国际研讨会论文集 77-。 86（1991）

Masaaki Shimasaki: "Compiling Technigucs for Supercomputers" Advances in Software Science and Technology. 4. 1-20 (1992)

Masaaki Shimasaki：“超级计算机的编译技术”软件科学与技术的进展。

館村純一 他: "マルチウィンドウデバッガHyper DEBUにおける細粒度高並列プログラムの実行のデータフローの視覚化" 情報処理学会論文誌. 34. 580-594 (1993)

Junichi Tatemura 等人：“在多窗口调试器 Hyper DEBU 中执行细粒度和高度并行程序时的数据流可视化”，日本信息处理学会汇刊 34. 580-594 (1993)。