Computer and Mathematical Sciences
Computer Architecture A12
Next-Generation High-Performance/Low-Power/Dependable Computer Architectures by Co-designing Hardware and Software
Our laboratory is active in the research on high-performance, low-power, and dependable computing systems by co-designing hardware and software environments. The details are as follows.
＜High-Performance/Low-Power/Dependable Computer Architecture Design＞
While computers have significantly increased their performances, they are also required to reliably process a large amount of data with the lower power. Our laboratory focuses on memory systems by using the 3D die-stacking and non-volatile devices, adaptive hardware technologies to maximize the performance per watt by providing resources to applications, and dependable technologies of computers.
＜Hybrid Processing Architectures of Machine Learning/Quantum Annealing and Conventional Computing, and their Applications with Social Deployments＞
Due to the power limitations of computers and the demands for big data processing, new computing methods such as machine learning (ML) and quantum annealing (QA) have attracted attention. Our laboratory is conducting researches on the integration of these emerging ML/QA technologies into the conventional computing as a single system image, and development and deployment of their killer applications: a real-time tsunami inundation simulation with routing assistance for evacuation, and a system to predict turbine faults.
Hybrid conventional-QA-AI/ML computing environment
A real-time tsunami inundation simulation with routing assistance for evacuation