The 265th R-CCS Cafe (July 1, 2024) | RIKEN Center for Computational Science RIKEN Website

TOP Events & Outreach R-CCS Cafe The 265th R-CCS Cafe (July 1, 2024)

Details
Date	Mon, Jul 1, 2024
Time	3:00 pm - 5:20 pm (3:00 pm - 4:00 pm Talks, 4:00 pm - 4:20 Discussions, 4:20 pm - Free discussion and coffee break)
City	Kobe, Japan/Online
Place	Lecture Hall (6th floor) at R-CCS, Online seminar on Zoom If you are not affiliated with R-CCS and would like to attend R-CCS Cafe, please email us at r-ccs-cafe[at]ml.riken.jp.
Language	Presentation Language: English Presentation Material: English
Speakers	Kentaro Sano Processor Research Team Team Leader Toshiyuki Imamura Large-scale Parallel Numerical Computing Technology Research Team Team Leader Miwako Tsuji Quantum-HPC Hybrid Software Environment Unit Unit Leader

Talk Titles and Abstracts

1st Speaker: Kentaro Sano

Title:
Architectures for future computing - from CGRA for HPC and AI to near-sensor/device custom computing -
Abstract:
At Processor research team and the next-generation AI device research-and-development unit, we research future architectures for HPC and AI. Especially, we are focusing on reconfigurable computing architectures such as coarse-grained reconfigurable array (CGRA), which can be advantageous for limited data movement resulting in lower power consumption, as well as FPGA-based custom computing for near-sensor/device processing. In this talk, I introduce our previous research on FPGA-based reconfigurable HPC and share lessons learned, and then show RIKEN CGRA project for HPC and AI, FPGA-based edge-computing projects for SPring-8 detector and fault-tolerant quantum computers.

2nd Speaker: Toshiyuki Imamura

Title:
The recent advanced research on numerical linear algebra
Abstract:
With its unwavering commitment, the large-scale parallel numerical computing technology research team has been at the forefront of establishing universal and portable HPC mathematical software for emerging systems on the earth. Our past achievements include the development of several original numerical libraries supported on Fugaku, such as EigenExa, FFTE-C, dissection, Batched BLAS generator, tmBLAS, and more. These accomplishments are a testament to our team's dedication and expertise in numerical linear algebra.
In our recent work, 'mixed precision' has emerged as a significant keyword. This approach, which involves considering multiple data formats simultaneously, is crucial for maximizing processors' potential and aligning with future architectural trends.
As we recognized the complexities of utilizing AI-driven or AI-empowered processors, also known as NPU, for numerical operations, we have encountered various challenges. However, our team's unwavering dedication and expertise have led us to develop innovative solutions. These include a multi-component data format for high-precision, double-double, or pair-arithmetic technologies. Ozaki's scheme is a testament to our team's and the Japanese HPC community's innovative work history. These solutions have shown promising results under high-performance and massively equipped low-precision arithmetic mechanisms such as NPUs, instilling confidence in our ability to overcome these challenges.
We have been conducting research activities focusing on these fundamental technologies. As part of our fundamental activities, I would like to discuss the prospects of mX_real and Ozaki's scheme.

3rd Speaker: Miwako Tsuji

Title:
Design and Development of Quantum HPC Hybrid Software Environment
Abstract:
Last year, which was a first year of the quantum HPC hybrid platform division, we had investigated requirements for the quantum HPC Hybrid software environment and designed our software. In this talk, I will introduce our "HPC-oriented" quantum HPC Hybrid software design and current states of their developments.

Important Notes

Please turn off your video and microphone when you join the meeting.
The broadcasting may be interrupted or terminated depending on the network condition or any other unexpected event.
The program schedule and contents may be modified without prior notice.
Depending on the utilized device and network environment, it may not be able to watch the session.
All rights concerning the broadcasted material will belong to the organizer and the presenters, and it is prohibited to copy, modify, or redistribute the total or a part of the broadcasted material without the previous permission of RIKEN.

（Jun 21, 2024）