●2021年12月23日 Osni Marques　特任教授（名古屋大学大学院情報学研究科／米国ローレンスバークレー国立研究所） 主催： 名古屋大学大学院情報学研究科　価値創造センター 名古屋大学大学院情報学研究科 題目： An Overview of the IDEAS-ECP Project and Minimization Strategies for Solving Eigenvalue Problem 講演者：Dr. Osni Marques　 概要：This presentation will consist of two parts. In the first part, I will give an overview of IDEAS-ECP, a project funded by the US Department of Energy Exascale Project (ECP) that seeks to promote software quality as essential for quality science. IDEAS-ECP evaluates and disseminates best practices and methodologies to improve developer productivity, software sustainability, and scientific reproducibility. I will summarize the activities carried by IDEAS-ECP, which range from collaborations with individual ECP teams to help in the understand productivity bottlenecks and improve software development practices, to interactions with ECP software developers to help identify training and support needs, and the organization of a variety of events that foster the discussion of software development experiences. In the second part, I will discuss ongoing work on unconstrained minimization schemes for dealing with eigenvalue problems. These schemes employ a preconditioned conjugate gradient approach that avoids an explicit reorthogonalization of the trial eigenvectors, in contrast to typical iterative eigenvalue solvers, therefore reducing communications and becoming a potential alternative for the solution of large problems on massively parallel computers. Our current focus is on problems related to electronic structure calculations. Preliminary results show that the unconstrained formulation, together with an appropriate preconditioner, offers good convergence properties and scales well on a large number of cores. 　〇当日発表資料 PDF ●2018年8月10日Franz Franchetti教授（米国カーネギーメロン大学） 主催：科学技術研究費補助金　基盤研究（B） 「通信回避・削減アルゴリズムのための自動チューニング技術の新展開」 日時： 　２０１８年８月１０日（金）１３：３０～１４：３０ 　名古屋大学高等総合研究館３階３０８セミナー室 　http://www.nagoya-u.ac.jp/access-map/higashiyama/iar.html 　　※地下鉄名城線名古屋大学駅出口から徒歩２０分程度かかりますので、ご注意ください。 参加費無料（事前登録不要ですが、会場のセキュリティのため事前連絡が要ります） Title: Formal Software Synthesis of Computational Kernels Presenter: Franz Franchetti, Professor, ECE Department, Carnegie Mellon University Abstract In this talk we address the question of how to automatically map computational kernels across a wide range of computing platforms to highly efficient code, and prove the correctness of the synthesized code. This addresses two fundamental problems that software developers are faced with: performance portability across the ever-changing landscape of parallel platforms, and verifiable correctness of sophisticated floating-point code. The problem is attacked as follows: We develop a formal framework to capture computational algorithms, computing platforms, and program transformations of interest, using a unifying mathematical formalism we call operator language (OL). Then we cast the problem of synthesizing highly optimized computational kernels for a given machine as a strongly constrained optimization problem that is solved by a multi-stage rewriting system. Since all rewrite steps are semantics preserving identity operations, our approach allows us to formally prove the equivalence between the kernel specification and the synthesized program. We have implemented this approach as part of the SPIRAL system where we have formalized a selection of computational kernels from the signal and image processing domain, software-defined radio, and robotic vehicle control. We have targeted platforms spanning from mobile devices as well as desktop and server multicore processors to large high performance and supercomputing systems. Our provably correct synthesized vehicle safety monitors and controllers have been demonstrated on an unmanned ground robot and a passenger car. Bio: Franz Franchetti is a Professor in the Department of Electrical and Computer Engineering at Carnegie Mellon University. He received the Dipl.-Ing. (M.Sc.) degree in Technical Mathematics and the Dr. techn. (Ph.D.) degree in Computational Mathematics from the Vienna University of Technology in 2000 and 2003, respectively. In 2006 he was member of the team winning the Gordon Bell Prize (Peak Performance Award) and in 2010 he was member of the team winning the HPC Challenge Class II Award (most productive system). Dr. Franchetti's research focuses on automatic performance tuning and program generation for emerging parallel platforms and algorithm/hardware co-synthesis. He leads/led three DARPA projects in the BRASS, HACMS and PERFECT programs, and is PI/Co-PI on a number of federal and industry grants. He is one of the principal designers of the SPIRAL system.

●2019年8月6日Franz Franchetti教授（米国カーネギーメロン大学） 主催：科学技術研究費補助金　挑戦的研究（萌芽） 「ディープラーニングを利用した革新的自動チューニング基盤の創製」 日時： 　２０１９年８月６日（水）１４：４５～１５：４５ 　名古屋大学情報基盤センター　４階セミナー室 Title: SPIRAL’s Operator Language: From Textbook Math to High Performance -With Correctness Guarantees This talk discusses the Operator Language, the enabling technology in the SPIRAL code generation and automatic performance tuning system. We develop a formal framework to capture computational algorithms, computing platforms, and program transformations of interest, using a unifying mathematical formalism we call operator language (OL). Then we cast the problem of synthesizing highly optimized computational kernels for a given machine as a strongly constrained optimization problem that is solved by a multi-stage rewriting system. Since all rewrite steps are semantics preserving identity operations, our approach allows us to formally prove the equivalence between the kernel specification and the synthesized program. Finally we present a first look at FFTX and SpectralPack. We aim at translating the LAPACK/BLAS approach from the numerical linear algebra world to the spectral algorithm domain. FFTX is extending and updating FFTW for the exascale era and beyond while providing backwards compatibility. SpectralPack captures higher level spectral algorithms and their variants, including convolutions, Poisson solvers, correlations, and numerical differentiation approaches that translate to FFT calls. The SPIRAL system--now available as open source under a BSD/Apache license--underpins the effort to provide performance portability. Bio: Franz Franchetti is a full Professor in the Department of Electrical and Computer Engineering at Carnegie Mellon University. He received the Dipl.-Ing. (M.Sc.) degree in Technical Mathematics and the Dr. techn.(Ph.D.) degree in Computational Mathematics from the Vienna University of Technology in 2000 and 2003, respectively. Dr. Franchetti's research focuses on automatic performance tuning and program generation for emerging parallel platforms and algorithm/hardware co-synthesis.  Within the Spiral effort, his research goal is to enable automatic generation of highly optimized software libraries for important kernel functionality.

●2019年8月9日　Osni Marques博士（米国ローレンスバークレー国立研究所） 主催：科学技術研究費補助金　挑戦的研究（萌芽） 「ディープラーニングを利用した革新的自動チューニング基盤の創製」 日時： 　２０１９年８月９日（金）１３：３０～１４：３０ 　名古屋大学情報基盤センター　４階セミナー室

最終更新：２０２１年１月１１日