本文へジャンプ

Let’s build high-performance computers

TANAKA Laboratory
Associate Professor:TANAKA Kiyofumi

E-mail:
[Research areas]
Computer Architecture, Real-Time Embedded Systems
[Keywords]
Memory system, Caches, energy reduction, FPGA, real-time scheduling, embedded OS

Skills and background we are looking for in prospective students

Our researches are based on knowledge of “Digital Logic and Computer Design”, ”Computer Architecture”, and “Operating Systems”. In most researches, sufficient skills for programming as well as enthusiasm for high performance and/or energy reduction are required.

What you can expect to learn in this laboratory

By carrying out researches, students are well-trained for various, important ability for investigation, problem finding, resolution, planning, implementation, evaluation, writing reports, and presentation, which are all required in the future career. In particular, from experience of designing computer hardware logic, they can learn to image concurrent behaviors of objects and realize them, leading to issue resolution.

【Job category of graduates】 Information/communication-processing industry

Research outline

In our life environment, computers are embedded everywhere, e.g., mobile phones, home appliances, networking equipment, and vehicles, where real-time property is often required. Along with growing demands for IoT devices, we study how to build systems with efficient real-time processing, from a viewpoint of both hardware and basic software. Especially, we design actual systems for empirical researches.

■ Real-time embedded systems

Real-Time Embedded RISC Processor
Embedded processors must have low cost and low energy consumption. In addition, response performance is important for control applications. We extend conventional RISC architecture and design real-time embedded RISC core with multi-contexts And efficient cache memory, which makes response performance much higher.

Real-Time Embedded Operating System
In real-time systems, high response time performance as well as meeting real-time constraints is important. To improve the response time, use of appropriate scheduling policies is indispensable. We develop real-time embedded operating systems which improve real-time processing as much as possible by taking dynamic behavior into account while conforming to the common API standard of embedded operating systems

■ Network-on-Chip

Networks-on-Chip (NoCs) are becoming increasingly important elements in different types of computing hardware platforms such as many-core processors and accelerators. In the near future, NoCs with hundreds to thousands of nodes will be required. However, a major obstacle to research and development of such large-scale NoCs is the lack of fast modeling methodologies that can provide a high degree of accuracy. We are tackling this problem using FPGAs.

■ Other researches

Accelerators for Deep Learning Hardware, Adaptation/configuration of processors/OS, system implementation on FPGA, low-energy consumption architecture, real-time resource management, virtual ECU for automatic driving, implementation of RISC-V processors, etc.

Key publications

  1. D.Doan, K.Tanaka, "A Novel Task-to-Processor Assignment Approach for Optimal Multiprocessor Real-time Scheduling," Proc of IEEE Intl. Symp. on Embedded Multicore/Many-core Systems-On-Chip (MCSoC), pp.101-108, 2018. (Best Paper Award)
  2. T. V. Chu, S. Sato, and K. Kise, “Fast and Cycle-Accurate Emulation of Large-Scale Networks-on-Chip Using a Single FPGA,” ACM Transactions on Reconfigurable Technology and Systems, Vol. 10, No. 4, pp.27.1-27.27, 2017.
  3. K. Tanaka, “Real-Time Scheduling for Reducing Jitters of Periodic Tasks,” Journal of Information Processing, Vol.23, No.5, pp.542-552, 2015.

Equipment

Linux/Solaris servers and storages
FPGA evaluation boards
FPGA development environment
CAD/Verification tools
Measuring equipment (oscilloscope, logic analyzer)

Teaching policy

We have a laboratory-local seminar once a week and one-on-one meeting/discussion every day. In the lab local meeting, all members deal with the common issues to learn to read literature, understand it, and explain the content in his or her own words. In the one-on-one meeting, the student and supervisor set a short-term goal of research and discuss about how to achieve it. The supervisor gives advices to students, while the students are supposed to solve each problem by their own.

[Website] URL:http://tlab-web.jaist.ac.jp/

PageTop