Much of the improvement in computer performance over the last 40 years has been provided by computer architecture advancements that have leveraged Moore's Law and Dennard scaling to build larger and more parallel systems. Moore's Law is the observation that the maximum number of transistors in an integrated circuit doubles approximately every two years. Dennard scaling refers to the reduction of MOS supply voltage in concert with the scaling of feature sizes, so that as transistors get smaller, their power density stays roughly constant. With the end of Dennard scaling a decade ago, and the recent slowdown of Moore's Law due to a combination of physical limitations and economic factors, the sixth edition of the preeminent textbook for our field couldn't be more timely. Here are some reasons.
First, because domain-specific architectures can provide equivalent performance and power benefits of three or more historical generations of Moore's Law and Dennard scaling, they now can provide better implementations than may ever be possible with future scaling of general-purpose architectures. And with the diverse application space of computers today, there are many potential areas for architectural innovation with domain-specific architectures. Second, high-quality implementations of open-source architectures now have a much longer lifetime due to the slowdown in Moore's Law. This gives them more opportunities for continued optimization and refinement, and hence makes them more attractive. Third, with the slowing of Moore's Law, different technology components have been scaling heterogeneously. Furthermore, new technologies such as 2.5D stacking, new nonvolatile memories, and optical interconnects have been developed to provide more than Moore's Law can supply alone. To use these new technologies and nonhomogeneous scaling effectively, fundamental design decisions need to be reexamined from first principles. Hence it is important for students, professors, and practitioners in the industry to be skilled in a wide range of both old and new architectural techniques. All told, I believe this is the most exciting time in computer architecture since the industrial exploitation of instruction-level parallelism in microprocessors 25 years ago.
The largest change in this edition is the addition of a new chapter on domain-specific architectures. It's long been known that customized domain-specific architectures can have higher performance, lower power, and require less silicon area than general-purpose processor implementations. However when general-purpose processors were increasing in single-threaded performance by 40% per year (see Fig. 1.11), the extra time to market required to develop a custom architecture vs. using a leading-edge standard microprocessor could cause the custom architecture to lose much of its advantage. In contrast, today single-core performance is improving very slowly, meaning that the benefits of custom architectures will not be made obsolete by general-purpose processors for a very long time, if ever. Chapter 7 covers several domain-specific architectures. Deep neural networks have very high computation requirements but lower data precision requirements – this combination can benefit significantly from custom architectures. Two example architectures and implementations for deep neural networks are presented: one optimized for inference and a second optimized for training. Image processing is another example domain; it also has high computation demands and benefits from lower-precision data types. Furthermore, since it is often found in mobile devices, the power savings from custom architectures are also very valuable. Finally, by nature of their reprogrammability, FPGA-based accelerators can be used to implement a variety of different domain-specific architectures on a single device. They also can benefit more irregular applications that are frequently updated, like accelerating internet search.
Although important concepts of architecture are timeless, this edition has been thoroughly updated with the latest technology developments, costs, examples, and references. Keeping pace with recent developments in open-sourced architecture, the instruction set architecture used in the book has been updated to use the RISC-V ISA.
On a personal note, after enjoying the privilege of working with John as a graduate student, I am now enjoying the privilege of working with Dave at Google. What an amazing duo!