Research Topics in Computer Architecture
Overall Course Objectives
The course will approach selected modern circuit and architecture design problems from an unusual low level perspective: Asynchronous circuit and system design. This course intends to teach the students to think different and out-of the box about digital and memory architectures. It will start off with the basics in asynchronous design and then follow up with selected major architecture challenges and how to think about them differently with a data-centric and data-driven computer architecture approach.
Computer architecture, the art and science of designing hardware, is an exciting and fast-changing research and development field. Participants are expected to work on a project with an oral presentation with appropriate presentation material during the 13-week period and deliver a final report.
See course description in Danish
Learning Objectives
- Evaluate and explain whether it is advantageous to use asynchronous circuits when implementing a digital circuit for a given application, and assess the effect on the properties of the circuit (energy, speed, area, etc.).
- Identify and explain the problems (synchronization and metastability) related to communicating between several clock domains, devise possible solutions and reason about the performance and reliability of these.
- Analyse and explain when data-centric architectures are advantages over compute-centric architectures on practical examples.
- Assess which handshake protocol is most appropriate to use in a given situation.
- Analyze performance and power trade-offs in modern computer architectures
- Design and implement designs in an asynchronous hardware description language.
- Evaluate the power consumption/latency based on data communication distance and deduce architectural implications out of the evaluation.
- Give a coherent oral presentation of a chosen topic, based on a project
- Prepare a conference-style paper (report)
Course Content
The topics considered in this course will change from year to year. Typical topics could include instruction-level parallelism, asynchronous circuits, chip-multiprocessing, network-on-chip, time-predictable computer architectures, AI accelerators, hardware for security, and approximate computing. The 2026/2027 topics are asynchronous circuit and architecture design and data-centric architectures. The specific plan for the year will be announced via an email to the course participants before the course starts.
Teaching Method
The course consists of lectures, exercises, and project work.
Faculty
Remarks
Students will work in groups of two or three on exercises/labs and problems to practice some topics covered in the lectures. Further students will prepare a final project/lab selected from a catalog of possible topics building on top of the topics covered in the course.




