This report summarizes the contents of the panoramic view workshop "Computing Architecture" held on June 12, 2024.

Moore's Law, which has brought about dramatic advances in computer performance, is showing signs of slowing down, and we are being forced to respond to demands for even greater performance and new workloads, such as the recent emergence of generative AI. In addition, there are growing demands for realtime operation in autonomous driving and expectations for edge computing. Furthermore, power saving along with the increase in processing speed has become an important issue on a global scale.

In light of this situation, we believe that a new computing architecture that breaks away from conventional thinking is needed, and that is why we planned this workshop. In this workshop, we will discuss new computing paradigms, the processors that realize them, and domain specific architectures. The purpose is to clarify the direction, research and development themes, and associated challenges that could serve as strategies to strengthen research and development in Japan in the future.

First, the six invited experts gave position talks on current research, noteworthy technologies, and trends in academic circles. This was followed by a general discussion, with the following points being discussed:

1. Collaboration between Device Science and Computer Science
   Devices and computing are not separate things; they work together and evolve together. Therefore, close cooperation between the two par ties is necessary. In addition to the division into devices and computing, there are also structures such as devices, circuits, architectures, and algorithms, or even the possibility of other new structure. In any case, collaboration across multiple levels is necessary. Funding plays an important role in promoting collaboration across multiple fields like this. Measures are needed to create a cohesive team, and team-building activities are necessary as a preliminary step.
2. The Importance of System Software
   New devices and computing alone won't make anything work. It may seem mundane, but it won't actually work without a compiler, OS, EDA design, etc. Peripheral technologies are important.
3. Improving energy efficiency
   The discussion at this workshop highlighted that much of computing ultimately aims to improve computational energy efficiency. In such situations, learning from the brain is one approach.
4. Continuity is key
   The number of architecture researchers is declining, and if things continue as they are, they may become extinct. Several related funding efforts have been made to date, but it is more important than ever that these efforts continue.

Based on the discussions at this workshop, we plan to continue our studies in order to make concrete proposals.