AIST started operation of ABCI, which is a publicly available cloud infrastructure for AI research and development, from August 2018, and RWBC-OIL uses ABCI for AI research. RWBC-OIL researchers from AIST and TokyoTech attend SC19 to provide presentations and booth exhibition about ABCI and AI/BigData research projects.

SC19 Booth Exhibition

Highlights at SC19 is presentations of RWBC-OIL research activities that use ABCI and TSUBAME3.0. This year, three research papers and three posters from AIST and TokyoTech are presented. There are also many talks from RWBC-OIL researchers at BoF, panels, workshop and exhibition booths.

Same as last year's SC, we present ABCI at the booth. ABCI is an open innovation platform with world-class computing resources of 550 Pflops (HP) / 37 Pflops (DP) for AI research and development. The design of ABCI was impacted by the design of TSUBAME3.0 at TokyoTech. Further information about ABCI is available from ABCI web site. Information about TSUBAME3.0 is available from TokyoTech web site. You can reach the exhibition information about TokyoTech on this page.

Booth Map

AIST booth is at #1117
TokyoTech booth is at #313

Presentations @ Conference

Technical Paper

HyperX Topology: First At-Scale Implementation and Comparison to the Fat-Tree
Jens Domke, Satoshi Matsuoka, Ivan Radanov Ivanov, Yuki Tsushima, Tomoya Yuki, Akihiro Nomura, Shin'ichi Miura, Nic McDonald, Dennis Lee Floyd, Nicolas Dubé
Date: Wednesday, 20 November 2019, 11am - 11:30am
Location:301-302-303

A Versatile Software Systolic Execution Model for GPU Memory Bound Kernels
Peng Chen, Mohamed Wahib, Shinichiro Takizawa, Ryousei Takano, Satoshi Matsuoka
Date: Wednesday, 20 November 2019, 4:30pm-5pm
Location: 405-406-407

iFDK: A Scalable Framework for Instant High-Resolution Image Reconstruction
Peng Chen, Mohamed Wahib, Shinichiro Takizawa, Ryousei Takano, Satoshi Matsuoka
Date: Thursday, 21 November 2019, 3:30pm - 4pm
Location:301-302-303

Research Poster

Poster 48: Runtime System for GPU-Based Hierarchical LU Factorization
Qianxiang Ma, Rio Yokota
Date: Thursday, 21 November 2019, 8:30am - 5pm
Location:E Concourse

Poster 96: TSQR on TensorCores
Hiroyuki Ootomo, Rio Yokota
Date: Thursday, 21 November 2019, 8:30am - 5pm
Location:E Concourse

Poster 111: Multiple HPC Environments-Aware Container Image Configuration for Bioinformatics Application
Kento Aoyama, Hiroki Watanabe, Masahito Ohue, Yutaka Akiyama
Date: Thursday, 21 November 2019, 8:30am - 5pm
Location:E Concourse

Panel

Early Career Program: Welcome and Leadership Panel/Discussion
Satoshi Matsuoka
Date: Monday, 18 November 2019, 8:30am - 10am
Location:709

Open Science in a Changing HPC Landscape
Jason Haga (on behalf of Satoshi Sekiguchi)
Date: Wednesday, 20 November 2019, 1:30pm-3pm
Location:201-203

HPC Big Data and AI: Computing under Constraints
Satoshi Matsuoka
Date: Friday, 22 November 2019, 10:30am-12pm
Location:205-207

BoF

Energy Efficiency Considerations for HPC Procurements
Ryousei Takano
Date: TimeTuesday, 19 November 201912:15pm - 1:15pm
Location:601-603

Workshop

On the Feasibility of Optical Circuit Switching for Distributed Deep Learning
Truong Thao Nguyen
Workshop: PHOTONICS
Date: Monday, 18 November 2019, 12:10pm - 12:30pm
Location:710

Panel: Rejuvenate Post-Moore's Law Information Systems with Photonics
Shu Namiki
Workshop: PHOTONICS
Date: Monday, 18 November 2019, 4:30pm - 5:30pm
Location:710

Presentations @ AIST Booth

SAGE2 Session
Jason Haga, Maxine Brown, Jason Leigh
Date: Tuesday, 19 November 2019, 2pm - 3:30pm

ABCI Application Session
Gerald Pao, Kyoung-Sook Kim, Katsuki Fujisawa
Date: Tuesday, 19 November 2019, 4pm - 5pm

SAGE2 BoF Redo
Jason Haga, Maxine Brown, Jason Leigh
Date: Wednesday, 20 November 2019, 11am - 12pm

Workshop @ AIST Booth

Title: Approximate Computing in the Post-Moore Era (Mini-Symposium)
Date: Tuesday, 19 November 2019, 10:15-12:00

Dr. Ryousei Takano from AIST (5 minutes opening remarks)
Dr. Takano/Hirofuchi/Wahib from AIST (10:20-10:40) Overview of the Approximate Project at AIST
Dr. Michael O. Lam from JMU (10:40-11:00) Automated Software Tools for Mixed Precision
Dr. Leonardo Bautista from BSC (11:00-11:20) Aggressive Undervolting of FPGAs: Power & Reliability Trade-offs
Dr. Ignacio Laguna from LLNL (11:20-11:40) Performance-Driven Floating-Point Tuning for GPU Applications
Dr. Ikki Fujiwara from NII (11:40-12:00) High-Bandwidth Low-Latency Approximate Interconnection Networks