R-CCS Cafe 第286回（2026年2月13日）

トップページイベント・広報 R-CCS Cafe R-CCS Cafe 第286回（2026年2月13日）

詳細
開催日	2026年2月13日(金)
開催時間	15:00 - 16:30（15:00 - 16:00　講演者3名による講演、16:00~　自由討論（参加自由））
開催都市	兵庫県神戸市／オンライン
場所	計算科学研究センター（R-CCS）6階講堂／Zoomによる遠隔セミナー R-CCS外部の方で参加希望の場合は r-ccs-cafe[at]ml.riken.jp までご連絡ください。
使用言語	発表・スライド共に英語
登壇者	吉岡直樹量子計算シミュレーション技術開発ユニット技師 Wenyang Zhao 計算構造生物学研究チーム特別研究員 Jiaheng Liu プロセッサ研究チーム特別研究員

講演題目・要旨

1st Speaker: Naoki Yoshioka

Title:

State-vector simulation of quantum computers on supercomputers

Abstract:

A state-vector simulator of quantum computers has been developed for use on supercomputers. Our simulator, RIKEN-braket, removes the conventional limitations on the number of MPI processes and the size of the state-vector data array imposed by commonly used parallelization methods. We demonstrate that our simulator scales efficiently up to 46 qubits on the supercomputer Fugaku, using up to 55,296 computing nodes. We also describe recent enhancements to the simulator, including gate fusion and support for multiple circuits to enable simulations of variational algorithms.

2nd Speaker: Wenyang Zhao

Title:

xspline3d: A Python Library for MPI-Based Spline Interpolation Enforcing Global Continuity in Distributed 3D Volumes

Abstract:

Spline interpolation plays a fundamental role in scientific computing and data science. In recent years, the size of datasets has been increasing rapidly, requiring them to be partitioned across distributed memory. However, performing multivariate spline interpolation on large-scale, distributed, multidimensional data is non-trivial, as it requires the entire system to be solved globally to enforce global continuity. To address this challenge, we propose parallel algorithms to solve the global multidimensional system dimension by dimension, while progressively redistributing the data across memory blocks in different MPI processes. We implement these algorithms as a Python library xspline3d, the first publicly available library supporting MPI-based multivariate spline interpolation in distributed 3D volumes. This library effectively overcomes the single-node memory bottleneck for handling large-scale datasets. It is expected to evolve into a widely adopted utility library on HPC platforms.

3rd Speaker: Jiaheng Liu

Title:

On-chip Scratchpad Memory Design Exploration for RIKEN CGRA SoC

Abstract:

Elastic coarse-grained reconfigurable arrays (CGRAs) use back-pressure mechanisms to provide autonomous, decentralized flow-control, avoiding static scheduling and strict assumptions on tile-level latency. This elastic characteristic also eases coupling with external modules that have dynamic, data-dependent latency, widening the design space of on-chip scratchpad memories (SPMs). We evaluate multiple SPM designs for the RIKEN CGRA SoC, focusing on power and area as well as their impact on throughput and data duplication across representative kernels. This work highlights the trade-off: fully-connected multibank SPM spends substantial power/area on interconnects, whereas private port–bank design can increase data duplication for kernels with shared working sets. Thus, we evaluate a reduced-connectivity multibank SPM in which each port accesses only a subset of banks. This design provides a more favorable performance-efficiency balance, making it practical for RIKEN CGRA SoC.

注意事項

参加の際はPCマイクの音声・ビデオをオフにされるようお願いいたします。
当日の会場環境や通信状態により、やむなく配信を中止・中断する場合がございます。
プログラムの内容、時間は予告なく変更される場合があります。
ご使用の機器やネットワークの環境によっては、ご視聴いただけない場合がございます。
インターネット中継に関する著作権は、主催者及び発表者に帰属します。なお、配信された映像及び音声、若しくはその内容を、理化学研究所の許可無くほかのウェブサイトや著作物等への転載、複製、改変等を行うことを禁じます。

（2026年2月10日）