RESEARCH/研究紹介

The Advanced Computing Systems Laboratory (ACSL) is a cutting-edge research facility dedicated to exploring and innovating in the field of advanced computing technologies. ACSL serves as a collaborative hub for students, faculty, and industry partners, focusing on the development of advanced AI-enabled software hardware architectures and systems for various applications. The mission of the laboratory is to drive research excellence in advanced computing systems, fostering innovation and knowledge transfer that addresses real-world challenges. We aim to enhance computational capabilities, improve system efficiencies, and contribute to the advancement of computing technologies that support diverse fields such as artificial intelligence, automotive edge, and scientific computing. We focus on a variety of key research areas, including:

Power/Energy-Efficient Computing Architectures and Systems/SoCs: Development of power-efficient architectures and algorithms that maximize computational speed and efficiency for scientific simulations and data analysis. 電力/エネルギー効率の高いコンピューティング アーキテクチャとシステム/SoC：科学的シミュレーションおよ びデータ解析のための計算速度と効率を最大化する電力効率の高いアーキテクチャとアルゴリ ズムの開発。

Brain-inspired Algorithms and Systems: Designing hardware and software that simulate neural and synaptic structures to process information in a way similar to biological systems. 脳にヒントを得たアルゴリズムとシステム：生物システムに似た方 法で情報を 処理するために神経およびシナプス構造をシミュレートするハードウェアおよびソフトウェア の

Advanced On-chip Interconnects: Focus on the communication pathways within a multicore chip, which are essential for connecting various components like processors, memory, and other functional units (2D/3D, Si-Photonics, Hybrid) 高度なオンチップ相互接続：プロセッサ、メモリ、その 他の機 能ユニット（2D/3D、Si-Photonics、ハイブリッド）などの様々なコンポー ネントを接続するために不可欠なマルチコアチップ内の通信経路に注力。

Automotive Edge Computing: Investigation of decentralized computing models that bring computation closer to data sources, improving response times and bandwidth efficiency in  future automotive. 自動車エッジコンピューティング：計算をデータ ソースに近づけること で応答 時間と帯域幅効率を改善する、将来の自動車における分散型コンピューティングモデルの調 査。