CS6640: Advanced Computer Architecture (3-1-0-0-8-12-12)

Objective:

The end of Moore's Law and Dennard scaling has prompted a search for alternative architectures and technologies to ensure systems remain capable, fast, energy-efficient, programmable, and reliable. This course begins with a brief overview of sequential computer architecture concepts, then shifts focus to parallel computer architectures. Next, the course explores the integration of specialized, heterogeneous components within parallel architectures. Finally, the course examines emerging models of computation and technologies.

Syllabus:

Brief Overview of Sequential Computer Architecture: Introduction, Superscalar processors, Multithreaded processors, and RISC-V open instruction set architecture.
Parallel Computer Architecture: Programmability, Parallel memory system design, Coherence, Consistency, Synchronisation, Transactional hardware, Hardware support for programming, debugging, and avoiding failures, Manycore designs, Interconnection networks, and Dark silicon.
Heterogeneous Computer Architecture: Domain-specific architectures, Data-Level Parallelism, CPU/FPGA hybrid systems, Big.LITTLE heterogeneous multicores, CPU/GPU hybrid systems, and Heterogeneous memory consistency issues.
Emerging Computer Architecture Concepts: Processing-in-memory and near-memory processing, Neuromorphic computing, Approximate computing, Novel memory technologies, 3D-stacking architectures, Chiplet architectures, and Graph-centric architecture.
Paper Presentations by Students

Reference Material:

J.L. Hennessy, D.A. Patterson, and C. Kozyrakis. Computer Architecture: A Quantitative Approach. 7th Edition, Elsevier Publishers, 2025.
D. Culler, J.P. Singh, and A. Gupta. Parallel Computer Architecture: A Hardware/Software Approach. Morgan Kaufmann Publishers, 1998.
Relevant Research Papers.

Evaluation Mechanism:

Lecture Venue and Schedule: CS26 and E Slot (Tue-11:00-11:50; Wed-10:00-10:50; Thu-08:00-08:50; Fri-17:00-17:50)

Last modified: Feb 26th, 2026.