HOME > 研究会・イベント > ASE研究会 > 第15回ASE研究会


 今回は、招待講演として フランスの Maison de la Simulation/CNRS および University Lille 1から Serge G. Petiton 教授をお呼びし、エクサスケールのスーパーコンピュータに向けた、クリロフ部分空間法ベースのアルゴリズムについての講演を行います。


日時:2013 年 10 月 23 日(水)14:55 ~ 16:45(懇親会:17:30 ~)
場所:東京大学 情報基盤センター(浅野地区)4階遠隔会議室(地図
主催:東京大学情報基盤センター スーパーコンピューティング部門



15th Advanced Supercomputing Environment (ASE) Seminar

October 23th (Wednesday), 2013
14:55 - 16:45

Information Technology Center (Asano Section), The University of Tokyo
4F Telecommunication Meeting Room(Map

14:55 - Opening

15:00 - 16:00 Invited Talk

Invited Speaker

Professor Serge G. Petiton
Maison de la Simulation/CNRS, and University Lille 1, Sciences et Technologies


On the road to Intelligent Krylov-based Methods for Extreme Computing


Exascale hypercomputers are expected to have highly hierarchical architectures with nodes composed by lot-of-core processors and accelerators. The different programming levels (from clusters of processors loosely connected to tightly connected lot-of-core processors and/or accelerators) will generate new difficult algorithm issues. New methods should be defined and evaluated with respect to modern state-of-the-art of applied mathematics and scientific methods.
Krylov linear methods such as GMRES and ERAM are now heavily used with success in various domains and industries despite their complexity. Their convergence and speed greatly depends on the hardware used and on the choice of the Krylov subspace size and other parameters which are difficult to determine efficiently in advance. Moreover, hybrid Krylov Methods would allow reducing the communications along all the cores, limiting the reduction only through subsets of these cores. Added to their numerical behaviors and their fault tolerance properties, these methods are interesting candidates for exascale/extreme matrix computing. Avoiding communication strategies may also be developed for each of the instance of these methods, generating complex methods but with high potential efficiencies. These methods have a lot of correlated parameter which may be optimized using auto/smart-tuning strategies to accelerate convergence, minimize storage space, data movements, and energy consumption.
In this talk, we first will present some basic matrix operations utilized on Krylov methods on clusters of accelerators, with respect to a few chosen sparse compressed formats. We will discuss some recent experiments on a cluster of accelerators concerning comparison between orthogonal, incompletely orthogonal and non-orthogonal Krylov Basis computing. Then, we will discuss some results obtained on a cluster of accelerators to compute eigenvalues using the MERAM method with respect to the restarting strategies. We will survey some auto/smart-tuning strategies we proposed and evaluated for some of the Krylov method parameters. As a conclusion, we will propose auto-tuning strategies for future hybrid methods on post-petascale computers, on the road to exascale intelligent hybrid methods.

16:00 - 16:15 Break

16:15 - 16:45 Regular Presentation


Takahiro Katagiri
Information Technology Center, The University of Tokyo


Towards Directive-base Auto-tuning Language for Supercomputers in Operation
--- Open Questions and Future Direction ---


Although several Auto-tuning (AT) languages have been proposed and studied for the last decade, we do not still have crucial tools and languages for AT for supercomputers in operation. In this presentation, we present another AT language based on directive-base, which is ppOpen-AT. With respect to supercomputer environments, preprocessor and AT framework of ppOpen-AT do not require any daemons and script languages for code generation and AT cycle phases.
As a scenario of AT, we adapt the AT function to a real application based on finite difference method (FDM). By using the Fujitsu PRIMEHPC FX10, that is supercomputer in operation in the University of Tokyo, we obtain essential performance improvement by adapting the AT function. We will also point out several open questions and future direction for directive-base AT languages.

16:45 - Closing

17:30 - Reception

( 10 月 20 日(日)12時までに、katagiri@cc.u-tokyo.ac.jpまで送付 )

 日時:2013 年 10 月 23 日(水)17:30 ~ @根津駅周辺
 予算:5,000 円程度


 備 考:領収書が必要な方はここにご記載ください。(宛名:     )


  • センターユーザに限定せず、研究会は一般公開とします。
  • 参加費は無料で、基本的に事前登録は不要です。



〒113-8658 東京都文京区弥生2-11-16
東京大学 情報基盤センター

ASE研究会幹事 准教授 片桐孝洋