Modern Computer Architecture and Organization (3rd Edition) (github.com)

The third edition of "Modern Computer Architecture and Organization" by Jim Ledin has been released, reflecting significant updates and expansions tailored to recent advancements in technology. The new edition introduces two critical chapters focused on Graphics Processing Units (GPUs) and Large Language Models (LLMs), driven by the explosive growth of AI and its impacts on computing architecture. These chapters analyze the architecture and processing capabilities of modern GPUs, detailing their dual roles in AI training and real-time inference. For the AI/ML community, this update is particularly significant as it addresses the evolving demands of large-scale computing infrastructures that support advanced AI operations. The analysis of GPT-2 serves as a case study to elucidate the processing requirements of contemporary LLMs, emphasizing the essential physical and operational needs of AI datacenters, including resource allocation for power and cooling. By including practical exercises and code examples, the book not only educates readers on theoretical concepts but also reinforces hands-on skills vital for navigating the current landscape of AI technology.