狠狠撸shows by User: takahirokatagiri58 / http://www.slideshare.net/images/logo.gif 狠狠撸shows by User: takahirokatagiri58 / Sat, 05 Oct 2024 12:26:37 GMT 狠狠撸Share feed for 狠狠撸shows by User: takahirokatagiri58 情报処理学会贬笔颁研究会贬笔颁196研究会パネル讨论导入部资料（2024年9月30日） /slideshow/hpc-hpc196-2024-9-30/272205794 hpc19620240930-katagiri-241005122637-a3586a48
This is the material for the panel discussion at the Information Processing Society of Japan, Special Interest Group, HPC, 196 (30th September, 2024).]]>
This is the material for the panel discussion at the Information Processing Society of Japan, Special Interest Group, HPC, 196 (30th September, 2024).]]> Sat, 05 Oct 2024 12:26:37 GMT /slideshow/hpc-hpc196-2024-9-30/272205794 takahirokatagiri58@slideshare.net(takahirokatagiri58) 情报処理学会贬笔颁研究会贬笔颁196研究会パネル讨论导入部资料（2024年9月30日） takahirokatagiri58 This is the material for the panel discussion at the Information Processing Society of Japan, Special Interest Group, HPC, 196 (30th September, 2024). <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/hpc19620240930-katagiri-241005122637-a3586a48-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> This is the material for the panel discussion at the Information Processing Society of Japan, Special Interest Group, HPC, 196 (30th September, 2024).
情报処理学会贬笔颁研究会贬笔颁196研究会パネル讨论导入部资料（2024年9月30日） from Takahiro Katagiri
]]> 11 0 https://cdn.slidesharecdn.com/ss_thumbnails/hpc19620240930-katagiri-241005122637-a3586a48-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Auto‐Tuning of Hierarchical Computations with ppOpen‐AT /slideshow/autotuning-of-hierarchical-computations-with-ppopenat/62495638 siampp16-hp-160528152937
We are now developing ppOpen-AT, which is a directive-base Auto-tuning (AT) language to specify fundamental AT functions, i.e., varying values of parameters, loop transformations, and code selection. Considering with expected hardware of Post Moore’s era, we focus on optimization for computations with deep hierarchy of 3D memory stack. ppOpen-AT provides code selection to optimize code with respect to layers of the memory. Performance evaluation of AT with a code of FDM will be shown by utilizing the Xeon Phi. ]]>
We are now developing ppOpen-AT, which is a directive-base Auto-tuning (AT) language to specify fundamental AT functions, i.e., varying values of parameters, loop transformations, and code selection. Considering with expected hardware of Post Moore’s era, we focus on optimization for computations with deep hierarchy of 3D memory stack. ppOpen-AT provides code selection to optimize code with respect to layers of the memory. Performance evaluation of AT with a code of FDM will be shown by utilizing the Xeon Phi. ]]> Sat, 28 May 2016 15:29:37 GMT /slideshow/autotuning-of-hierarchical-computations-with-ppopenat/62495638 takahirokatagiri58@slideshare.net(takahirokatagiri58) Auto‐Tuning of Hierarchical Computations with ppOpen‐AT takahirokatagiri58 We are now developing ppOpen-AT, which is a directive-base Auto-tuning (AT) language to specify fundamental AT functions, i.e., varying values of parameters, loop transformations, and code selection. Considering with expected hardware of Post Moore’s era, we focus on optimization for computations with deep hierarchy of 3D memory stack. ppOpen-AT provides code selection to optimize code with respect to layers of the memory. Performance evaluation of AT with a code of FDM will be shown by utilizing the Xeon Phi. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/siampp16-hp-160528152937-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> We are now developing ppOpen-AT, which is a directive-base Auto-tuning (AT) language to specify fundamental AT functions, i.e., varying values of parameters, loop transformations, and code selection. Considering with expected hardware of Post Moore’s era, we focus on optimization for computations with deep hierarchy of 3D memory stack. ppOpen-AT provides code selection to optimize code with respect to layers of the memory. Performance evaluation of AT with a code of FDM will be shown by utilizing the Xeon Phi.
Auto‐Tuning of Hierarchical Computations with ppOpen‐AT from Takahiro Katagiri
]]> 360 5 https://cdn.slidesharecdn.com/ss_thumbnails/siampp16-hp-160528152937-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 辫辫翱辫别苍-贬笔颁コードを自动チューニングする言语辫辫翱辫别苍-础罢の现状と新展开 /takahirokatagiri58/ppopenhpcppopenat atta2015-at-hp-151225065334
本発表では、ppOpen-HPCプロジェクトによる実シミュレーションコードを自動チューニング（AT）する言語ppOpen-ATの現状報告を行う。今後の展開として、多階層なメモリなどを想定し、多階層にATを実施する場合の例を、有限差分法（FDM）によるアプリケーションppOpen-APPL/FDMを題材にしたAT適用例を紹介する。また、Xeon Phiでのチューニング結果も紹介する。]]>
本発表では、ppOpen-HPCプロジェクトによる実シミュレーションコードを自動チューニング（AT）する言語ppOpen-ATの現状報告を行う。今後の展開として、多階層なメモリなどを想定し、多階層にATを実施する場合の例を、有限差分法（FDM）によるアプリケーションppOpen-APPL/FDMを題材にしたAT適用例を紹介する。また、Xeon Phiでのチューニング結果も紹介する。]]> Fri, 25 Dec 2015 06:53:34 GMT /takahirokatagiri58/ppopenhpcppopenat takahirokatagiri58@slideshare.net(takahirokatagiri58) 辫辫翱辫别苍-贬笔颁コードを自动チューニングする言语辫辫翱辫别苍-础罢の现状と新展开 takahirokatagiri58 本発表では、ppOpen-HPCプロジェクトによる実シミュレーションコードを自動チューニング（AT）する言語ppOpen-ATの現状報告を行う。今後の展開として、多階層なメモリなどを想定し、多階層にATを実施する場合の例を、有限差分法（FDM）によるアプリケーションppOpen-APPL/FDMを題材にしたAT適用例を紹介する。また、Xeon Phiでのチューニング結果も紹介する。 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/atta2015-at-hp-151225065334-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> 本発表では、ppOpen-HPCプロジェクトによる実シミュレーションコードを自動チューニング（AT）する言語ppOpen-ATの現状報告を行う。今後の展開として、多階層なメモリなどを想定し、多階層にATを実施する場合の例を、有限差分法（FDM）によるアプリケーションppOpen-APPL/FDMを題材にしたAT適用例を紹介する。また、Xeon Phiでのチューニング結果も紹介する。
辫辫翱辫别苍-贬笔颁コードを自动チューニングする言语辫辫翱辫别苍-础罢の现状と新展开 from Takahiro Katagiri
]]> 1185 5 https://cdn.slidesharecdn.com/ss_thumbnails/atta2015-at-hp-151225065334-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Towards Automatic Code Selection with ppOpen-AT: A Case of FDM - Variants of Numerical Computations and ?Its Impact on a Multi-core Processor - /slideshow/towards-automatic-code-selection-with-ppopenat-a-case-of-fdm-variants-of-numerical-computations-and-its-impact-on-a-multicore-processor-55784982/55784982 spsn201520151203-hp-151203150944-lva1-app6892
In this study, we show a new ability of auto-tuning (AT) by utilizing selection of code variants based on totally different implementations of numerical computations. The selection function of the AT is carefully designed to apply ppOpen-AT, which is a computer language to adapt AT functions to simulation codes of actual use in ppOpen-HPC project. The AT is evaluated with ppOpen-APPL/FDM (Seism_3D), which is a simulation code of seismic wave based on Finite Difference Method (FDM). According to results of performance evaluation with an advanced multi-core processor, the Xeon Phi, crucial speedups are found by utilizing the selection of AT. Moreover, the best code variants were varied according to parallel executions, i.e. the number of MPI processes and OpenMP threads in hybrid MPI/OpenMP. ]]>
In this study, we show a new ability of auto-tuning (AT) by utilizing selection of code variants based on totally different implementations of numerical computations. The selection function of the AT is carefully designed to apply ppOpen-AT, which is a computer language to adapt AT functions to simulation codes of actual use in ppOpen-HPC project. The AT is evaluated with ppOpen-APPL/FDM (Seism_3D), which is a simulation code of seismic wave based on Finite Difference Method (FDM). According to results of performance evaluation with an advanced multi-core processor, the Xeon Phi, crucial speedups are found by utilizing the selection of AT. Moreover, the best code variants were varied according to parallel executions, i.e. the number of MPI processes and OpenMP threads in hybrid MPI/OpenMP. ]]> Thu, 03 Dec 2015 15:09:44 GMT /slideshow/towards-automatic-code-selection-with-ppopenat-a-case-of-fdm-variants-of-numerical-computations-and-its-impact-on-a-multicore-processor-55784982/55784982 takahirokatagiri58@slideshare.net(takahirokatagiri58) Towards Automatic Code Selection with ppOpen-AT: A Case of FDM - Variants of Numerical Computations and ?Its Impact on a Multi-core Processor - takahirokatagiri58 In this study, we show a new ability of auto-tuning (AT) by utilizing selection of code variants based on totally different implementations of numerical computations. The selection function of the AT is carefully designed to apply ppOpen-AT, which is a computer language to adapt AT functions to simulation codes of actual use in ppOpen-HPC project. The AT is evaluated with ppOpen-APPL/FDM (Seism_3D), which is a simulation code of seismic wave based on Finite Difference Method (FDM). According to results of performance evaluation with an advanced multi-core processor, the Xeon Phi, crucial speedups are found by utilizing the selection of AT. Moreover, the best code variants were varied according to parallel executions, i.e. the number of MPI processes and OpenMP threads in hybrid MPI/OpenMP. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/spsn201520151203-hp-151203150944-lva1-app6892-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> In this study, we show a new ability of auto-tuning (AT) by utilizing selection of code variants based on totally different implementations of numerical computations. The selection function of the AT is carefully designed to apply ppOpen-AT, which is a computer language to adapt AT functions to simulation codes of actual use in ppOpen-HPC project. The AT is evaluated with ppOpen-APPL/FDM (Seism_3D), which is a simulation code of seismic wave based on Finite Difference Method (FDM). According to results of performance evaluation with an advanced multi-core processor, the Xeon Phi, crucial speedups are found by utilizing the selection of AT. Moreover, the best code variants were varied according to parallel executions, i.e. the number of MPI processes and OpenMP threads in hybrid MPI/OpenMP.
Towards Automatic Code Selection with ppOpen-AT: A Case of FDM - Variants of Numerical Computations and Its Impact on a Multi-core Processor - from Takahiro Katagiri
]]> 537 6 https://cdn.slidesharecdn.com/ss_thumbnails/spsn201520151203-hp-151203150944-lva1-app6892-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 ソフトウェア自动チューニング研究绍介 /slideshow/ss-54760598/54760598 researchoverview20151105-rev1-hp-151105024156-lva1-app6891
本発表では、ソフトウェア自动チューニング分野の研究概要について绍介し、着者らが関係している最新の话题のごく一部について、绍介します。闭闭>
本発表では、ソフトウェア自动チューニング分野の研究概要について绍介し、着者らが関係している最新の话题のごく一部について、绍介します。闭闭> Thu, 05 Nov 2015 02:41:56 GMT /slideshow/ss-54760598/54760598 takahirokatagiri58@slideshare.net(takahirokatagiri58) ソフトウェア自动チューニング研究绍介 takahirokatagiri58 本発表では、ソフトウェア自动チューニング分野の研究概要について绍介し、着者らが関係している最新の话题のごく一部について、绍介します。 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/researchoverview20151105-rev1-hp-151105024156-lva1-app6891-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> 本発表では、ソフトウェア自动チューニング分野の研究概要について绍介し、着者らが関係している最新の话题のごく一部について、绍介します。
ソフトウェア自动チューニング研究绍介 from Takahiro Katagiri
]]> 1587 8 https://cdn.slidesharecdn.com/ss_thumbnails/researchoverview20151105-rev1-hp-151105024156-lva1-app6891-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Ase20 20151016 hp /slideshow/ase20-20151016-hp/54005635 ase2020151016hp-151016045812-lva1-app6891
In this research, we show effect of auto-tuning (AT) for function of code selection to computational kernels for scientific and technology computations. ppOpen-AT, which is a computer language to specify AT function to arbitrary parts of program, is utilized to describe the code selection. The evaluation of AT in this research performed with advanced CPU architectures, such as the Intel Xeon Phi and the Intel Ivy Bridge. Results of preliminary experiment with a code based on Finite Difference Method (FDM) indicate that the effect of AT is crucial with compared to conventional AT framework without code selection.]]>
In this research, we show effect of auto-tuning (AT) for function of code selection to computational kernels for scientific and technology computations. ppOpen-AT, which is a computer language to specify AT function to arbitrary parts of program, is utilized to describe the code selection. The evaluation of AT in this research performed with advanced CPU architectures, such as the Intel Xeon Phi and the Intel Ivy Bridge. Results of preliminary experiment with a code based on Finite Difference Method (FDM) indicate that the effect of AT is crucial with compared to conventional AT framework without code selection.]]> Fri, 16 Oct 2015 04:58:12 GMT /slideshow/ase20-20151016-hp/54005635 takahirokatagiri58@slideshare.net(takahirokatagiri58) Ase20 20151016 hp takahirokatagiri58 In this research, we show effect of auto-tuning (AT) for function of code selection to computational kernels for scientific and technology computations. ppOpen-AT, which is a computer language to specify AT function to arbitrary parts of program, is utilized to describe the code selection. The evaluation of AT in this research performed with advanced CPU architectures, such as the Intel Xeon Phi and the Intel Ivy Bridge. Results of preliminary experiment with a code based on Finite Difference Method (FDM) indicate that the effect of AT is crucial with compared to conventional AT framework without code selection. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/ase2020151016hp-151016045812-lva1-app6891-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> In this research, we show effect of auto-tuning (AT) for function of code selection to computational kernels for scientific and technology computations. ppOpen-AT, which is a computer language to specify AT function to arbitrary parts of program, is utilized to describe the code selection. The evaluation of AT in this research performed with advanced CPU architectures, such as the Intel Xeon Phi and the Intel Ivy Bridge. Results of preliminary experiment with a code based on Finite Difference Method (FDM) indicate that the effect of AT is crucial with compared to conventional AT framework without code selection.
Ase20 20151016 hp from Takahiro Katagiri
]]> 581 5 https://cdn.slidesharecdn.com/ss_thumbnails/ase2020151016hp-151016045812-lva1-app6891-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 辫辫翱辫别苍-础罢による静的コード生成で実现する?自动チューニング方式の评価 /slideshow/ppopenat/52652877 jsiam2015-katagiri-hp-150911010039-lva1-app6891
本報告では、静的コード生成のみによる自動チューニング(Auto-tuning、AT)の実現方式であるStatic Code Generation Auto-tuning (SCG-AT)を提案し、実アプリケーションによる有効性検証を行う。また、SCG-ATによるATを行う際に、最も効率的であると思われる「階層型AT処理」を実装し、コード選択のATとして実現した。]]>
本報告では、静的コード生成のみによる自動チューニング(Auto-tuning、AT)の実現方式であるStatic Code Generation Auto-tuning (SCG-AT)を提案し、実アプリケーションによる有効性検証を行う。また、SCG-ATによるATを行う際に、最も効率的であると思われる「階層型AT処理」を実装し、コード選択のATとして実現した。]]> Fri, 11 Sep 2015 01:00:39 GMT /slideshow/ppopenat/52652877 takahirokatagiri58@slideshare.net(takahirokatagiri58) 辫辫翱辫别苍-础罢による静的コード生成で実现する?自动チューニング方式の评価 takahirokatagiri58 本報告では、静的コード生成のみによる自動チューニング(Auto-tuning、AT)の実現方式であるStatic Code Generation Auto-tuning (SCG-AT)を提案し、実アプリケーションによる有効性検証を行う。また、SCG-ATによるATを行う際に、最も効率的であると思われる「階層型AT処理」を実装し、コード選択のATとして実現した。 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/jsiam2015-katagiri-hp-150911010039-lva1-app6891-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> 本報告では、静的コード生成のみによる自動チューニング(Auto-tuning、AT)の実現方式であるStatic Code Generation Auto-tuning (SCG-AT)を提案し、実アプリケーションによる有効性検証を行う。また、SCG-ATによるATを行う際に、最も効率的であると思われる「階層型AT処理」を実装し、コード選択のATとして実現した。
ppOpen-ATによる静的コード生成で実現する 自動チューニング方式の評価 from Takahiro Katagiri
]]> 825 8 https://cdn.slidesharecdn.com/ss_thumbnails/jsiam2015-katagiri-hp-150911010039-lva1-app6891-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 厂颁骋-础罢：静的コード生成のみによる自动チューニング実现方式 /slideshow/scgat/51328662 hpc150-rev0-hp-150806015725-lva1-app6892
本報告では，自動チューニング（AT）を実行するに当たり，コード最適化時に動的なコード生成とコンパイルを行わず，実行前に静的に生成したコードのみを利用するATソフトウェア構成方式のStatic Code Generation Auto-tuning (SCG-AT)を提案する．SCG-ATによるATを評価するにあたり「階層型AT処理」を実装した．差分法による地震波シミュレーションppOpen-APPL/FDMにおいて，従来のベクトル計算機向けコードと新規開発したスカラ計算機向けコードのコード選択処理を実装した．Xeon Phi，Ivy Bridge，およびFX10の3種の全く異なる計算機でSSG-ATによるコード選択のATを評価した．評価の結果，Xeon PhiとIvy Bridgeにおいてはスカラ計算機向けコードの選択により，従来行われていたAT方式では達成できない速度向上が達成できることを明らかにした． ------ ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。 Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof. All Rights Reserved, Copyright (C) Information Processing Society of Japan. Comments are welcome. Mail to address editj＠ipsj.or.jp, please.]]>
本報告では，自動チューニング（AT）を実行するに当たり，コード最適化時に動的なコード生成とコンパイルを行わず，実行前に静的に生成したコードのみを利用するATソフトウェア構成方式のStatic Code Generation Auto-tuning (SCG-AT)を提案する．SCG-ATによるATを評価するにあたり「階層型AT処理」を実装した．差分法による地震波シミュレーションppOpen-APPL/FDMにおいて，従来のベクトル計算機向けコードと新規開発したスカラ計算機向けコードのコード選択処理を実装した．Xeon Phi，Ivy Bridge，およびFX10の3種の全く異なる計算機でSSG-ATによるコード選択のATを評価した．評価の結果，Xeon PhiとIvy Bridgeにおいてはスカラ計算機向けコードの選択により，従来行われていたAT方式では達成できない速度向上が達成できることを明らかにした． ------ ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。 Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof. All Rights Reserved, Copyright (C) Information Processing Society of Japan. Comments are welcome. Mail to address editj＠ipsj.or.jp, please.]]> Thu, 06 Aug 2015 01:57:25 GMT /slideshow/scgat/51328662 takahirokatagiri58@slideshare.net(takahirokatagiri58) 厂颁骋-础罢：静的コード生成のみによる自动チューニング実现方式 takahirokatagiri58 本報告では，自動チューニング（AT）を実行するに当たり，コード最適化時に動的なコード生成とコンパイルを行わず，実行前に静的に生成したコードのみを利用するATソフトウェア構成方式のStatic Code Generation Auto-tuning (SCG-AT)を提案する．SCG-ATによるATを評価するにあたり「階層型AT処理」を実装した．差分法による地震波シミュレーションppOpen-APPL/FDMにおいて，従来のベクトル計算機向けコードと新規開発したスカラ計算機向けコードのコード選択処理を実装した．Xeon Phi，Ivy Bridge，およびFX10の3種の全く異なる計算機でSSG-ATによるコード選択のATを評価した．評価の結果，Xeon PhiとIvy Bridgeにおいてはスカラ計算機向けコードの選択により，従来行われていたAT方式では達成できない速度向上が達成できることを明らかにした． ------ ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。 Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof. All Rights Reserved, Copyright (C) Information Processing Society of Japan. Comments are welcome. Mail to address editj＠ipsj.or.jp, please. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/hpc150-rev0-hp-150806015725-lva1-app6892-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> 本報告では，自動チューニング（AT）を実行するに当たり，コード最適化時に動的なコード生成とコンパイルを行わず，実行前に静的に生成したコードのみを利用するATソフトウェア構成方式のStatic Code Generation Auto-tuning (SCG-AT)を提案する．SCG-ATによるATを評価するにあたり「階層型AT処理」を実装した．差分法による地震波シミュレーションppOpen-APPL/FDMにおいて，従来のベクトル計算機向けコードと新規開発したスカラ計算機向けコードのコード選択処理を実装した．Xeon Phi，Ivy Bridge，およびFX10の3種の全く異なる計算機でSSG-ATによるコード選択のATを評価した．評価の結果，Xeon PhiとIvy Bridgeにおいてはスカラ計算機向けコードの選択により，従来行われていたAT方式では達成できない速度向上が達成できることを明らかにした． ------ ここに掲載した著作物の利用に関する注意 本著作物の著作権は情報処理学会に帰属します。本著作物は著作権者である情報処理学会の許可のもとに掲載するものです。ご利用に当たっては「著作権法」ならびに「情報処理学会倫理綱領」に従うことをお願いいたします。 Notice for the use of this material The copyright of this material is retained by the Information Processing Society of Japan (IPSJ). This material is published on this web site with the agreement of the author (s) and the IPSJ. Please be complied with Copyright Law of Japan and the Code of Ethics of the IPSJ if any users wish to reproduce, make derivative work, distribute or make available to the public any part or whole thereof. All Rights Reserved, Copyright (C) Information Processing Society of Japan. Comments are welcome. Mail to address editj＠ipsj.or.jp, please.
厂颁骋-础罢：静的コード生成のみによる自动チューニング実现方式 from Takahiro Katagiri
]]> 656 6 https://cdn.slidesharecdn.com/ss_thumbnails/hpc150-rev0-hp-150806015725-lva1-app6892-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 自动チューニングとビックデータ：机械学习の适用の可能性 /slideshow/a-tand-bigdata201507hp2/50837648 atandbigdata201507-hp2-150723091447-lva1-app6891
本発表では、数値計算におけるソフトウェア自動チューニング（Software Auto-tuning、AT）研究において、ビックデータを取り扱う可能性と、性能チューニングの自動化に機械学習を適用している国内外の従来研究について述べる。また、著者らが扱っているATの問題に対して、機械学習を適用した研究推進への期待を示す。]]>
本発表では、数値計算におけるソフトウェア自動チューニング（Software Auto-tuning、AT）研究において、ビックデータを取り扱う可能性と、性能チューニングの自動化に機械学習を適用している国内外の従来研究について述べる。また、著者らが扱っているATの問題に対して、機械学習を適用した研究推進への期待を示す。]]> Thu, 23 Jul 2015 09:14:47 GMT /slideshow/a-tand-bigdata201507hp2/50837648 takahirokatagiri58@slideshare.net(takahirokatagiri58) 自动チューニングとビックデータ：机械学习の适用の可能性 takahirokatagiri58 本発表では、数値計算におけるソフトウェア自動チューニング（Software Auto-tuning、AT）研究において、ビックデータを取り扱う可能性と、性能チューニングの自動化に機械学習を適用している国内外の従来研究について述べる。また、著者らが扱っているATの問題に対して、機械学習を適用した研究推進への期待を示す。 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/atandbigdata201507-hp2-150723091447-lva1-app6891-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> 本発表では、数値計算におけるソフトウェア自動チューニング（Software Auto-tuning、AT）研究において、ビックデータを取り扱う可能性と、性能チューニングの自動化に機械学習を適用している国内外の従来研究について述べる。また、著者らが扱っているATの問題に対して、機械学習を適用した研究推進への期待を示す。
自动チューニングとビックデータ：机械学习の适用の可能性 from Takahiro Katagiri
]]> 1084 5 https://cdn.slidesharecdn.com/ss_thumbnails/atandbigdata201507-hp2-150723091447-lva1-app6891-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 iWAPT2015_katagiri /slideshow/i-wapt2015-hp/49118230 iwapt2015-hp-150608114214-lva1-app6892
iWAPT2015, Hyderabad International Convention Centre, Hyderabad, INDIA, Session 2, 13:30 ‐14:00 May 29th, 2015]]>
iWAPT2015, Hyderabad International Convention Centre, Hyderabad, INDIA, Session 2, 13:30 ‐14:00 May 29th, 2015]]> Mon, 08 Jun 2015 11:42:14 GMT /slideshow/i-wapt2015-hp/49118230 takahirokatagiri58@slideshare.net(takahirokatagiri58) iWAPT2015_katagiri takahirokatagiri58 iWAPT2015, Hyderabad International Convention Centre, Hyderabad, INDIA, Session 2, 13:30 ‐14:00 May 29th, 2015 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/iwapt2015-hp-150608114214-lva1-app6892-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> iWAPT2015, Hyderabad International Convention Centre, Hyderabad, INDIA, Session 2, 13:30 ‐14:00 May 29th, 2015
iWAPT2015_katagiri from Takahiro Katagiri
]]> 549 1 https://cdn.slidesharecdn.com/ss_thumbnails/iwapt2015-hp-150608114214-lva1-app6892-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Towards Auto‐tuning for the Finite Difference Method in Era of 200+ Thread Parallelisms /slideshow/acsi2015-hp/44093208 acsi2015hp-150130180520-conversion-gate01
Oral Presentation at ACSI2015, Tsukuba, Application Session, January 28th (Wed) 9:45 ‐ 10:15. This is not publishing material.]]>
Oral Presentation at ACSI2015, Tsukuba, Application Session, January 28th (Wed) 9:45 ‐ 10:15. This is not publishing material.]]> Fri, 30 Jan 2015 18:05:20 GMT /slideshow/acsi2015-hp/44093208 takahirokatagiri58@slideshare.net(takahirokatagiri58) Towards Auto‐tuning for the Finite Difference Method in Era of 200+ Thread Parallelisms takahirokatagiri58 Oral Presentation at ACSI2015, Tsukuba, Application Session, January 28th (Wed) 9:45 ‐ 10:15. This is not publishing material. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/acsi2015hp-150130180520-conversion-gate01-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> Oral Presentation at ACSI2015, Tsukuba, Application Session, January 28th (Wed) 9:45 ‐ 10:15. This is not publishing material.
Towards Auto‐tuning for the Finite Difference Method in Era of 200+ Thread Parallelisms from Takahiro Katagiri
]]> 658 1 https://cdn.slidesharecdn.com/ss_thumbnails/acsi2015hp-150130180520-conversion-gate01-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 ppOpen-AT : Yet Another Directive-base AT Language /slideshow/dagstuhl2013-t/43024871 dagstuhl2013-t-141226161629-conversion-gate01
29. September bis 4. Oktober 2013, Dagstuhl Seminar 13401, Automatic Application Tuning for HPC Architectures, Session: infrastructures, 10:30-11:00, October 1st (TUE) , 2013.]]>
29. September bis 4. Oktober 2013, Dagstuhl Seminar 13401, Automatic Application Tuning for HPC Architectures, Session: infrastructures, 10:30-11:00, October 1st (TUE) , 2013.]]> Fri, 26 Dec 2014 16:16:29 GMT /slideshow/dagstuhl2013-t/43024871 takahirokatagiri58@slideshare.net(takahirokatagiri58) ppOpen-AT : Yet Another Directive-base AT Language takahirokatagiri58 29. September bis 4. Oktober 2013, Dagstuhl Seminar 13401, Automatic Application Tuning for HPC Architectures, Session: infrastructures, 10:30-11:00, October 1st (TUE) , 2013. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/dagstuhl2013-t-141226161629-conversion-gate01-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> 29. September bis 4. Oktober 2013, Dagstuhl Seminar 13401, Automatic Application Tuning for HPC Architectures, Session: infrastructures, 10:30-11:00, October 1st (TUE) , 2013.
ppOpen-AT : Yet Another Directive-base AT Language from Takahiro Katagiri
]]> 481 4 https://cdn.slidesharecdn.com/ss_thumbnails/dagstuhl2013-t-141226161629-conversion-gate01-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Impact of Auto-tuning of Kernel Loop Transformation by using ppOpen-AT /slideshow/spns2013-20131205/43024855 spns201320131205-141226161434-conversion-gate01
SPNS2013, December 5th -6th, 2013, Conference Room, 3F, Bldg.1, Earthquake Research Institute (ERI), The University of Tokyo, December 6th, 2013, ppOpen-HPC and Automatic Tuning (Chair: Hideyuki Jitsumoto), 1330-1400]]>
SPNS2013, December 5th -6th, 2013, Conference Room, 3F, Bldg.1, Earthquake Research Institute (ERI), The University of Tokyo, December 6th, 2013, ppOpen-HPC and Automatic Tuning (Chair: Hideyuki Jitsumoto), 1330-1400]]> Fri, 26 Dec 2014 16:14:34 GMT /slideshow/spns2013-20131205/43024855 takahirokatagiri58@slideshare.net(takahirokatagiri58) Impact of Auto-tuning of Kernel Loop Transformation by using ppOpen-AT takahirokatagiri58 SPNS2013, December 5th -6th, 2013, Conference Room, 3F, Bldg.1, Earthquake Research Institute (ERI), The University of Tokyo, December 6th, 2013, ppOpen-HPC and Automatic Tuning (Chair: Hideyuki Jitsumoto), 1330-1400 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/spns201320131205-141226161434-conversion-gate01-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> SPNS2013, December 5th -6th, 2013, Conference Room, 3F, Bldg.1, Earthquake Research Institute (ERI), The University of Tokyo, December 6th, 2013, ppOpen-HPC and Automatic Tuning (Chair: Hideyuki Jitsumoto), 1330-1400
Impact of Auto-tuning of Kernel Loop Transformation by using ppOpen-AT from Takahiro Katagiri
]]> 420 3 https://cdn.slidesharecdn.com/ss_thumbnails/spns201320131205-141226161434-conversion-gate01-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Towards Auto-tuning Facilities into Supercomputers in Operation - The FIBER approach and minimizing software-stack requirements - /slideshow/atat201403-hp/43024830 atat201403-hp-141226161135-conversion-gate02
2014 ATAT in HPSC, National Taiwan University, March 15, 2014 (Saturday), Performance 10:10-10:30]]>
2014 ATAT in HPSC, National Taiwan University, March 15, 2014 (Saturday), Performance 10:10-10:30]]> Fri, 26 Dec 2014 16:11:35 GMT /slideshow/atat201403-hp/43024830 takahirokatagiri58@slideshare.net(takahirokatagiri58) Towards Auto-tuning Facilities into Supercomputers in Operation - The FIBER approach and minimizing software-stack requirements - takahirokatagiri58 2014 ATAT in HPSC, National Taiwan University, March 15, 2014 (Saturday), Performance 10:10-10:30 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/atat201403-hp-141226161135-conversion-gate02-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> 2014 ATAT in HPSC, National Taiwan University, March 15, 2014 (Saturday), Performance 10:10-10:30
Towards Auto-tuning Facilities into Supercomputers in Operation - The FIBER approach and minimizing software-stack requirements - from Takahiro Katagiri
]]> 421 3 https://cdn.slidesharecdn.com/ss_thumbnails/atat201403-hp-141226161135-conversion-gate02-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 础罢罢础2014基盘叠导入（片桐） /slideshow/atta2014-20141225d/42996300 atta2014-20141225-d-141224193536-conversion-gate01
平成26年度の科研費基盤B「実行時自動チューニング機能付き疎行列反復解法ライブラリのエクサスケール化」の プロジェクト概要 に関する説明です。]]>
平成26年度の科研費基盤B「実行時自動チューニング機能付き疎行列反復解法ライブラリのエクサスケール化」の プロジェクト概要 に関する説明です。]]> Wed, 24 Dec 2014 19:35:36 GMT /slideshow/atta2014-20141225d/42996300 takahirokatagiri58@slideshare.net(takahirokatagiri58) 础罢罢础2014基盘叠导入（片桐） takahirokatagiri58 平成26年度の科研費基盤B?「実行時自動チューニング機能付き疎行列反復解法ライブラリの?エクサスケール化」の プロジェクト概要 に関する説明です。 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/atta2014-20141225-d-141224193536-conversion-gate01-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> 平成26年度の科研費基盤B?「実行時自動チューニング機能付き疎行列反復解法ライブラリの?エクサスケール化」の プロジェクト概要 に関する説明です。
础罢罢础2014基盘叠导入（片桐） from Takahiro Katagiri
]]> 1685 2 https://cdn.slidesharecdn.com/ss_thumbnails/atta2014-20141225-d-141224193536-conversion-gate01-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Overview of ppOpen-AT/Static for ppOpen-APPL/FDM ver. 0.2.0 /slideshow/pp-open-atfdmad20140730/42779193 ppopen-at-fdm-ad-20140730-141216201014-conversion-gate01
This is a material for overview of ppOpen-AT/Static for ppOpen-APPL/FDM ver. 0.2.0, which is numerical simulation software of a seismic wave analysis with function of automatic performance tuning (AT). Project of ppOpen-HPC is developing and supporting for this software. The effect of AT is shown with respect to several recent computer environments, such as multi-core (Ivy Bridge) and many-core (Xeon Phi).]]>
This is a material for overview of ppOpen-AT/Static for ppOpen-APPL/FDM ver. 0.2.0, which is numerical simulation software of a seismic wave analysis with function of automatic performance tuning (AT). Project of ppOpen-HPC is developing and supporting for this software. The effect of AT is shown with respect to several recent computer environments, such as multi-core (Ivy Bridge) and many-core (Xeon Phi).]]> Tue, 16 Dec 2014 20:10:14 GMT /slideshow/pp-open-atfdmad20140730/42779193 takahirokatagiri58@slideshare.net(takahirokatagiri58) Overview of ppOpen-AT/Static for ppOpen-APPL/FDM ver. 0.2.0 takahirokatagiri58 This is a material for overview of ppOpen-AT/Static for ppOpen-APPL/FDM ver. 0.2.0, which is numerical simulation software of a seismic wave analysis with function of automatic performance tuning (AT). Project of ppOpen-HPC is developing and supporting for this software. The effect of AT is shown with respect to several recent computer environments, such as multi-core (Ivy Bridge) and many-core (Xeon Phi). <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/ppopen-at-fdm-ad-20140730-141216201014-conversion-gate01-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> This is a material for overview of ppOpen-AT/Static for ppOpen-APPL/FDM ver. 0.2.0, which is numerical simulation software of a seismic wave analysis with function of automatic performance tuning (AT). Project of ppOpen-HPC is developing and supporting for this software. The effect of AT is shown with respect to several recent computer environments, such as multi-core (Ivy Bridge) and many-core (Xeon Phi).
Overview of ppOpen-AT/Static for ppOpen-APPL/FDM ver. 0.2.0 from Takahiro Katagiri
]]> 380 2 https://cdn.slidesharecdn.com/ss_thumbnails/ppopen-at-fdm-ad-20140730-141216201014-conversion-gate01-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 Extreme‐Scale Parallel Symmetric Eigensolver for Very Small‐Size Matrices Using A Communication-Avoiding for Pivot Vectors /takahirokatagiri58/pp14-201402-katagiri pp14201402katagiri-141216192104-conversion-gate01
We have developed a parallel eigensolver for very small-size matrices. Unlike conventional solvers, our design policy focusses on nature of non-blocking computations and reduced communications. A communication-avoiding approach for Householder pivot vectors is used to implement part of Householder inverse transformation. In addition to that, we implement some techniques for reducing communications by using non-blocking communications in tridiagonalization part. Performance of the solver with full nodes in the Fujitsu FX10 (76,800 cores) is also presented.]]>
We have developed a parallel eigensolver for very small-size matrices. Unlike conventional solvers, our design policy focusses on nature of non-blocking computations and reduced communications. A communication-avoiding approach for Householder pivot vectors is used to implement part of Householder inverse transformation. In addition to that, we implement some techniques for reducing communications by using non-blocking communications in tridiagonalization part. Performance of the solver with full nodes in the Fujitsu FX10 (76,800 cores) is also presented.]]> Tue, 16 Dec 2014 19:21:04 GMT /takahirokatagiri58/pp14-201402-katagiri takahirokatagiri58@slideshare.net(takahirokatagiri58) Extreme‐Scale Parallel Symmetric Eigensolver for Very Small‐Size Matrices Using A Communication-Avoiding for Pivot Vectors takahirokatagiri58 We have developed a parallel eigensolver for very small-size matrices. Unlike conventional solvers, our design policy focusses on nature of non-blocking computations and reduced communications. A communication-avoiding approach for Householder pivot vectors is used to implement part of Householder inverse transformation. In addition to that, we implement some techniques for reducing communications by using non-blocking communications in tridiagonalization part. Performance of the solver with full nodes in the Fujitsu FX10 (76,800 cores) is also presented. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/pp14201402katagiri-141216192104-conversion-gate01-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> We have developed a parallel eigensolver for very small-size matrices. Unlike conventional solvers, our design policy focusses on nature of non-blocking computations and reduced communications. A communication-avoiding approach for Householder pivot vectors is used to implement part of Householder inverse transformation. In addition to that, we implement some techniques for reducing communications by using non-blocking communications in tridiagonalization part. Performance of the solver with full nodes in the Fujitsu FX10 (76,800 cores) is also presented.
Extreme‐Scale Parallel Symmetric Eigensolver for Very Small‐Size Matrices Using A Communication-Avoiding for Pivot Vectors from Takahiro Katagiri
]]> 1225 2 https://cdn.slidesharecdn.com/ss_thumbnails/pp14201402katagiri-141216192104-conversion-gate01-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post http://activitystrea.ms/schema/1.0/posted 0 https://public.slidesharecdn.com/v2/images/profile-picture.png www.abc-lib.org/MyHTML/index.html https://cdn.slidesharecdn.com/ss_thumbnails/hpc19620240930-katagiri-241005122637-a3586a48-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/hpc-hpc196-2024-9-30/272205794 情报処理学会贬笔颁研究会贬笔颁196研究会パ... https://cdn.slidesharecdn.com/ss_thumbnails/siampp16-hp-160528152937-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/autotuning-of-hierarchical-computations-with-ppopenat/62495638 Auto‐Tuning of Hierarc... https://cdn.slidesharecdn.com/ss_thumbnails/atta2015-at-hp-151225065334-thumbnail.jpg?width=320&height=320&fit=bounds takahirokatagiri58/ppopenhpcppopenat 辫辫翱辫别苍-贬笔颁コードを自动チューニング...