Fundamentals of Computer Architecture

Abstract

Computer architecture is the science & art of designing and optimizing hardware components and the hardware/software interface to create a computer that meets design goals. This course covers modern processors architectures, state-of-the-art accelerators (e.g., GPUs, systolic arrays, ML accelerators), memory systems (caches, main memory, storage), with a focus on both fundamentals and research.

Objective

The major goal of this course is to equip students with the knowledge of the hardware architectures and hardware/software interfaces of modern computing platforms, including how they are designed and optimized, so that students can learn tradeoffs in design and how to make and evaluate such tradeoffs. A secondary goal is to equip students with the capability to do research and development in computer architecture. We aim to cover components of modern high-performance processor architectures (e.g., superscalar, multi-threaded, out-of-order, VLIW), state-of-the-art accelerators (e.g., systolic arrays, GPUs, data-centric architectures) for data-intensive applications (e.g., machine learning, graph analytics, bioinformatics), and memory systems (e.g., main memory, storage, and emerging memory technologies), etc. We will focus on fundamentals as well as cutting-edge research.

Content

The principles presented in the lecture are reinforced in the laboratory exercises using state-of-the-art research simulators and frameworks (e.g., Ramulator 2.0, MQSim, DRAM Bender) for more in-depth understanding of specific system components (e.g., processor, memory scheduling, prefetching, storage subsystem, memory controller, DRAM chips)

Resources