ECGR 6090: Special Topics: Interconnection Networks for Modern Multicore Systems and Accelerators

University of North Carolina, Charlotte

Directed study of current topics of special interest. Section information text: This course provides an in-depth, systems-oriented introduction to interconnection networks that serve as the backbone of modern computing platforms, including multicore processors, GPUs, cloud-scale systems, and emerging chiplet-based AI accelerators. The course focuses on Network-on-Chip (NoC) fundamentals and extends to modern interconnect challenges driven by AI/ML workloads, accelerator-rich architectures, and advanced packaging (2.5D/3D integration and chiplets). Topics include interconnect design goals and metrics; topology design and analysis; routing algorithms and protocols (including deadlock/livelock considerations); switching and flow control; router microarchitecture and implementation tradeoffs; congestion management and quality-of-service; and case studies spanning CPU NoCs, GPU fabrics, TPU/accelerator interconnects, and chiplet/packaged systems. Students will also learn and apply simulation tools (e.g., BookSim, GEM5), read and discuss recent research papers, and understand how interconnect choices impact performance, efficiency, scalability, and reliability in modern systems.

Average GPA: 3.62

Grade distribution records: 707 students across 21 terms.

Grade distribution

GradeStudentsPercent
A45864.8%
B19427.4%
C355.0%
W111.6%

Based on 707 student grade records across 21 terms and 19 professors.

Instructors

Still loading. This can take a moment on a slow connection.

Loading My Class Grades