Customer Reviews

Inside the Machine: An Illustrated Introduction to Microprocessors and Computer Architecture

5 star:		(13)
4 star:		(4)
3 star:		(2)
2 star:		(0)
1 star:		(0)

 

 

This book is an introduction to computers that fills the gap between classic and challenging books like Hennesy and Patterson's, and the large number of "How Your Computer Works" books that are too basic for engineers.

The first four chapters lay the conceptual groundwork for later chapters' studies of real-world microprocessors. These chapters use a simplified example processor, the DLW, to illustrate basic and intermediate concepts like the instructions/data distinction, assembly language programming, superscalar execution, pipelining, the programming model, and machine language. This section is essential reading for those who are new to the world of microprocessors.

The middle section of the book consists of detailed studies of two popular desktop processor lines: the Pentium line from Intel and the PowerPC from IBM and Motorola. These chapters walk the reader through the chronological development of each processor line, describing the evolution of the microarchitectures and instruction set architectures under discussion. Along the way, more advanced concepts such as speculative execution, vector processing, and instruction set translation are introduced and explored via a discussion of one or more real microprocessors. Throughout the middle part of the book the approach is to explain each new processor's features in terms of how they differ from analogous features found in predecessors or competitors. The comparative part of the book culminates in chapters 7 and 8 which consists of detailed comparisons of two starkly different and important processors: Intel's Pentium 4 and Motorola's MPC7450, popularly known as the G4e.

After a brief introduction to 64-bit computing and the 64-bit extensions to the popular x86 instruction set architecture in chapter 9, the microarchitecture of the first mass-market 64-bit processor, the IBM PowerPC 970, is treated in chapter 10. The study of the 970, the majority of which is also directly applicable to IBM's POWER4 mainframe processor, concludes the book's coverage of PowerPC processors. Chapter 11 covers the organization and functioning of the memory hierarchy found in almost all modern computers.

The final chapter contains a detailed examination of the latest generation of processors from Intel: the Pentium M, Core Duo and Core 2 Duo. This chapter contains the most detailed discussion of these processors available online or in print, and it contains some new information not previously released and specially granted by Intel for printing in this book.

I found this book a great read - it is both accessible and enlightening, even for someone with many years experience of working with microprocessors. I really liked how the author used 4-color diagrams to illustrate whatever point he was trying to make. I also appreciate that unlike Patterson's classic book, you don't need to review your combinatorics and discrete math before and after opening the book. The following is the table of contents:

Chapter 1: Basic Computing Concepts
Chapter 2: The Mechanics of Program Execution
Chapter 3: Pipelined Execution
Chapter 4: Superscalar Execution
Chapter 5: The Intel Pentium and Pentium Pro
Chapter 6: PowerPC Processors: 600 Series, 700 Series, and 7400
Chapter 7: Intel's Pentium 4 vs. Motorola's G4e: Approaches and Design Philosophies
Chapter 8: Intel's Pentium 4 vs. Motorola's G4e: The Back End
Chapter 9: 64-Bit Computing and x86-64
Chapter 10: The G5: IBM's PowerPC 970
Chapter 11: Understanding Caching and Performance
Chapter 12: Intel's Pentium M, Core Duo, and Core 2 Duo

It's possible to say you know how your computer works. But do you really know how your microprocessor does what it does? Without forcing you to take a crash course in engineering, Jon Stokes does a great job in uncovering the mysteries in the book Inside the Machine: An Illustrated Introduction to Microprocessors and Computer Architecture. It's an excellent read if you want to know what happens after you press the power button...

Contents: Basic Computing Concepts; The Mechanics of Program Execution; Pipelined Execution; Superscalar Execution; The Intel Pentium and Pentium Pro; PowerPC Processors - 600 Series, 700 Series, and 7400; Intel's Pentium 4 vs. Motorola's G4E - Approaches and Design Philosophies; Intel's Pentium 4 vs. Motorola's G4E - The Back End; 64-Bit Computing and x86-64; The G5 - IBM's PowerPC 970; Understanding Caching and Performance; Intel's Pentium M, Core Duo, and Core 2 Duo; Bibliography and Suggested Reading; Index

Normally, books like this are endless pages of painfully detailed descriptions of technology that only a true engineering geek could understand and love. For the rest of us mere mortals, we have to make do with simplistic descriptions of the chip that runs our computers. All the details are taken on faith. Stokes successfully bridges the gap between textbook details and real-life analogies to make the work of the CPU understandable. He starts off with the basics of how a CPU works and how instructions are executed. From there, he introduces the concept of pipelined instructions, and shows how that creates a much faster chip. But there are drawbacks, and when you're done reading you readily understand those limitations. Once the general groundwork is in place, the discussion moves to specific microprocessors in the market and how they are designed. Yes, those chips are highly complex, but Stokes lays a solid foundation that makes it possible to actually grasp what's going on without a Masters in chip design. By the time you're done with the book, you are well-equipped to understand why a 2.8 GHz processor may be infinitely faster than a 3.2 GHz processor, depending on how the design was implemented. The graphic illustrations are colorful and clear, and coupled with a conversational teaching tone, this book is... dare I say... "fun" to read.

Definitely a recommended read for anyone who wants to delve into microprocessor design without taking a four year degree program prior to doing so.

Jon Stokes' Inside the Machine falls somewhere between Computer Science textbook and Popular Science reading. It's packed with a lot of information that is very technical, while not quite going to the technical depth of a classroom textbook. It does make heavy use of analogy to render some hard to grasp concepts a bit easier for the non CS major.

Inside the Machine is fairly dense with both content and color. Lots of information is available here with colorful diagrams and illustrations to back it up. You'll need a more than basic understanding of computers and at least a bit of programming experience under your belt to get the most out of it. With that, the average computer enthusiast can pick up this book and find themselves in possession of a clear and concise guide to basic processor theory and real processor architecture. if you are interested in how microprocessors really work and why they were developed as they were but not interested in obtaining a CS degree this book is your first, best stop.

The book is divided into 12 chapters, with a bibliography and index following. The first four chapters lead the reader through basic computing concepts, discussing how a program actually executes when it arrives at the processor and brings us through pipelined and superscalar execution, ways to increase speed and throughput of processors.

Once a basic understanding of how the microprocessor works is reached, Stokes then disects a number of popular processors that have existed in the last decade and a half. These chapters cover the Intel's Pentium and Pentium Pro, the 600. 700 and 7400 Power PC processors, Intel's P4 vs Motorola's G4, 64-bit and x86-64 processors, the G5 and IBM's PowerPC 970 and finally Intel's Pentium M, Core Duo and Core 2 Duo processors. If you do want to delve deeper into the world of microprocessors the bibliography supplied in the book is a great resource.

If anything in that last paragraph sounded really interesting to you or made you say "Oh I loved that processor!" than this book should go on your to-read list.

I was not a CS major and though I do work in IT I'm not an expert on microprocessor architecture. I knew what I needed to know about the products to do my job. When this book came along it was a real joy for me to read it. I've looked at a number of popular text books concerned with microprocessors and while reading this book won't get you a job with Intel it will certainly be a bit easier to digest and thus offer a lot more to readers like myself.

Don't expect to breeze through this if you're going to pick it up though. You should glance through it first if you have a chance. If everything you're reading in the first four chapters is causing you to say "uh huh" and "oh yeah" then you're ready for some more advanced material and won't really need this unless you're into historical data about microprocessors. If you're familiar with some of the concepts and would like a solid grounding in current architecture based on understanding popular prior models then this book is certainly for you.

For me a broad understanding of how these things worked and a bit of programming was all I needed to enjoy this book and what it has to offer. While you won't be up late a few nights glued to the page (well, okay I was up late one night. I was really digging into the differences between the P4 and the G4 and thinking "Finally, that's what all this meant" when my wife told me in no uncertain terms to shut the light out and go to bed. That's just me though.) You will find just about every page full of useful theory or practical knowledge that will increase your understanding and prepare you for the next section of the book until you find yourself reading about processors being sold right now.

First let me say that, with some hesitation, this book is worth the read. If you're looking for a easy to understand text on x86 and RISC computing, this is your book. But after reading it I was left with more criticism than accolades. So let's start with the good:

From top to bottom it describes the microarchitecture of every Intel processor from the Pentium to the Core 2 Duo and likewise with every PowerPC processor with the exception of the G6, (which was released around the same time as this book). Jon Stokes (the writer) uses easily identifiable analogies to describe otherwise incredibly complex systems. He has a sort of "down to earth" way about his writing and throws in a few cute jokes about Oprah and smoking cigarettes. Anyone with a cursory knowledge of computing can understand 90% of the text in this book. All you need is basic logic skills, math, and a halfway decent memory for what was written in the chapter before. If you're looking for a text on how the Core 2 Duo works and what differentiates it from the Pentium 4 (or any other previous microarchitecture), this is your book. Same goes for the PowerPC line of CPU's. There's also GREAT chapters on caching and x86-64.

Now the bad:

The biggest bone to pick is that he almost completely leaves AMD out of the book. The only mention AMD gets is in regards to x86-64. He sort of gives AMD "cred" for x86-64 and then moves on to how it works. This is unfortunate because AMD has made a lot meaningful contributions to x86 computing; the most notable being the IMC on the Opteron. The IMC gets one sentence in the entire book, which is this: it exists. To expand upon this particular bone, he doesn't even mention Sun's SPARC or DEC's Alpha, both of which have had major influences on x86 and RISC. I was flat out floored that the IMC received so little mention in the caching chapter of the book.

Granted, AMD and Intel's x86 processors aren't radically different from one another (especially compared to PowerPC), but with the level of detail this book goes into, he could have at least dedicated a single chapter to the Opteron. Furthermore, it seems like he compares Intel processors to PowerPC/Gx processors just for comparisons sake. Now that PowerPC is essentially dead, why write half the book about it? Let's get real here, PowerPC isn't going to contribute anything meaningful to computing from here on out. It'll live on in supercomputing and very high end servers, but the aim of this book is desktop computing. Another major microarch that goes unexplored is Itanium. Yes, Itanium hasn't been a commercial success, but it's definitely the best performing 64 bit solution out there and it's a complete and total departure from x86 which makes it interesting in my book. (Yes, I do realize I just contradicted myself w/ the G6 and Itanium lines)

Moving on there's little mention of dual core computing or parallelism outside C2D. He puts a small paragraph in on virtualization, but doesn't explain what the benefits are. He also completely leaves out hyperthreading (which will reappearing in Penryn). He doesn't really explain the difference in manufacturing technologies or other parts of a computer's "system." You get the idea (in the VERY last paragraph) that the industry is moving towards parallelism and multicore computing, but he never explains why or what the benefits are. And my final complaint is that the book ends rather abruptly. There's no closing chapter or conclusion... just a small paragraph that says something to the effect of "intel is moving away from higher clock speeds and focusing on more cores and it's a really big deal." And BAM! it's over.

To sum it up, I was a little disappointed with the lack of AMD's mention and overall Jon doesn't really exude "excitement" about computing. He seems to forget he's writing a book and not a white paper. Still, the book is probably one of the more definitive publishings on Intel's modern day x86 CPU's and IBM/Motorola's PPC line of CPU's, all written in a way easily understood. I wouldn't say it was a complete waste of my time, but a newcomer to computing would walk away from this book with almost no knowledge on Intel's biggest (and most important) competitor: AMD. Jon makes out Intel's biggest competitor to be IBM and the PowerPC line of processors, which of course is absurd.

I hate to call this book biased, but it's hard not to after reading it.

Let me say that I wish I could give this book 4 1/2 stars. It's just shy of 5 stars, but I couldn't place this book alongside some of my favorite 5-star books of all time. Still, I really enjoyed reading Inside the Machine -- it's a great book that will answer many questions for the devoted technical reader.

Inside the Machine benefits from several strong features. First, the book's color illustrations are a treat, nicely explaining many topics. Second, the comparative approach taken by author Jon Stokes is a powerful and enlightening educational tool. By comparing aspects of different processors (G4e as "wide and shallow" vs P4 as "narrow and deep") I learned more than reading about the processors individually. Third, the author "builds" processors feature-by-feature, starting with the hypothetical DLW-1, continuing with the DLW-2, and then showing how his constructs compare against real processors. Modern processors are very complicated, and a powerful way to learn how they work is to start simple and progress from there. Fourth, astute readers might use Inside the Machine as a simple introduction to assembly language. The book doesn't teach assembly, but it shows, instruction by instruction, how it maps to machine language (bit by bit).

A few minor aspects of Inside the Machine caused me to not give the book five stars. First, I would have liked at least some coverage of the 386 and 486, prior to the Pentium. The i386 at least seems to be the least common denominator for many Unix variants, even though some now use the Pentium for that role. Second, some of the material seemed a little unorganized. For example, a chapter on caching (ch 11) appears to have been thrown after ch 10, but doesn't seem to fit there. Elsewhere, I have to wait to ch 10 to learn about the front side bus, in the middle of a discussion of the G5. I would have also liked to have learned a little more terminology associated with motherboards and the like. Finally, each chapter lacks a summary or conclusion. A few times I felt like the chapter just ended full-stop, with no sense of what had been discussed.

I think addressing these shortcomings would make for an excellent second edition. Perhaps including an appendix with a processor summary would help. I guess if I really want more details, I can turn to Scott Mueller's Upgrading and Repairing PCs; the 18th edition arrives this summer. Inside the Machine is the book that will help you understand how the components of a modern microprocessor function.

I received this book with a lot of apprehension; after all, I dropped the introductory CS course in college because it was a little too overwhelming for my business-major brain. My fears were unfounded, and as someone who works in training, I can also appreciate the care with which the concepts of microprocessors and computer programming are explained and developed in a logical fashion. Although I've read Jon's articles over the years, the added and expanded material in the book helped me to connect the ideas that I previously didn't have a firm grasp on. The text and examples begin on an introductory level, but still manage to go into considerable depth without losing their readability.

But what really impressed me as an instructional designer was the fact that it could not only function as a textbook in a college-level course, but also as a standalone reference. In evaluating course materials, one needs to consider its reference capacity and one's ability to learn via self-study from it. The reference value of this book for students will remain extremely helpful after they have used it in a course, and their ability to teach themselves from it will be critically important during the class. Because of the nature of textbooks, which focus primarily on teaching, the clarification and explanation of intermediary steps can be quite tedious and decrease the book's reference value. However, Jon does an excellent job and is one of the few who deftly balances these two needs; I was, and remain, impressed with "Inside the Machine," both as a learner and an instructor.

There are already numerous books on microprocessors. But of all those I have read, this comes across as one of the most elegant in the explanations of concepts. In part, it was greatly aided by the many colour diagrams scattered throughout the text. Colour was used to enhance the pedagogic utility of the diagrams. This vividness helps a reader grasp the underlying concepts.

To be sure, many people working in the field already know the book's ideas, and learnt these from earlier texts, which typically had fewer diagrams, and those in black and white. Which meant that, in part, you were smart enough to grasp those ideas. But maybe you had to work a little harder to reach that understanding, due to a paucity of diagrams, and poorly drawn ones at that. Whereas in this book, such ideas like pipelining are eloquently accompanied by good graphics.

I started with very limited understanding of computer architecture. What I had was a jumble of marketing terms I had picked up from reading popular PC magazines. This book helped me understand the basics of computer architecture, and quickly took me on a fun and insightful tour of major 32-bit and 64-bit architectures.
I have gained enough understanding from this book to move on to more quantitative treatments of processor design.

If you've ever wondered how your processor/computer works, this book is a great introduction. It starts off with the fundamentals (ALU, FPU, Superscalar, Pipelining, Instruction Sets, MicroArchitectures) and then proceeds to show specific implementations (Pentium, PowerPC, etc...). This book is modern, comprehensive, and accessible. A must have for tech buffs.

Inside the machine by Jon Stokes is a wonderful introductory treatment of computer and microprocessor architecture. Author started off with microprocessor concept as simple computing and calculating machine. Next he gradually moved into advanced topic such as pipelining and superscalar execution. Then he discusses Intel Pentium and Pentium architecture and PowerPC processor followed by comparison between Intel Pentium 4 and Motorola G4E. Different caching scheme are discussed towards the end of the book. The book closes with discussion of Intel latest line of processor such as Pentium M, Core Duo and Core 2 Duo.

Chapter 1: Basic Computing Concept

In this chapter, heart of computer--microprocessor--is discussed with a simple file-clerk model. Basically, microprocessor consists of a CPU, some storage, data and instruction bus. The author put a great stress on CPU functionality as a device which read, modify and write data. So a computer is like a file clerk who sits at his desk all day waiting for messages from his boss. Eventually, boss send a message for him to perform a task or calculation and where to get information from his personal file cabinet. So clerk first get the number, then perform the required task or calculation and finally put the results back in his file cabinet. I think the above analogy is very power to understanding the CPU role in modern computing. In essence, CPU functionality is very simple and every other system in the computer supports the CPU to accomplish this very task of read, modify and write task endlessly.

Chapter 2: The Mechanics of Program Execution

This chapter introduces the machine language, the programming model, ISA (Instruction Set Architecture), branch instruction and fetch-execute loop. Machine language structure such as opcodes, binary encoding of arithmetic and memory access instruction is discussed for hypothetical machine DLW-1. Basically, DLW-1 has four simple instructions: add, sub, load, store. The author shows binary encoding of these instructions. For example, the format for the machine language encoding of a register-type arithmetic instruction has following format: mode (1bit), opcode (2 bits), source1 (2 bits), source 2 (2 bits), destination (2 bits) and bit 10 to 15 are all zeros. Branch and conditional instructions are covered toward the end. This chapter actually gets into mechanics of how computer represent data and instruction and work toward the goal of the computer to actually read, modify and write data.

Chapter 3: Pipelined Execution

Pipelining is a technique that computers use to speed up the execution of the program. Pipeline is explained with a simple yet powerful SUV factory analogy. Five stages of this supposed factory are presented as: chassis construction, engine placement, doors and covering mounting, wheel assembly and paint. The idea is that each stage takes an hour to finish so that while one SUV is going through the pipeline other workers are not doing anything. In order to speed up the process, a new SUV can be sent when the first SUV is in the second stage; therefore, it will speed up the process and factory output. Because of this simple analogy, I finally understand how the pipelining works in the computer system. Pipeline stalls are also discussed toward the end of the chapter because of that pipeline's average instruction throughput reduces.

Chapter 4: Superscalar Execution

This chapter continues the theme from the last chapter to improve the efficiency of microprocessor and introduces another ALU on the chip. As an example, DLW-1 is extended to DLW-2; it has now two ALU, so it's able to execute two arithmetic instructions in parallel, it's called two-way superscalar. The instruction "Decode" is relabeled as "Decode/Dispatch". Superscalar computing allows a microprocessor to increase the number of instruction per clock and can have multiple instructions write stage on each clock cycle. Toward the end of the chapter various challenges to superscalar design are discussed. Conditions under which two arithmetic instructions cannot be safely dispatch in parallel for simultaneous execution are called hazards. There are three different categories: data hazards, structure hazard and control hazards. Overall, this chapter provided good overview of pipelining and superscalar deign and it's challenges to implement them in microprocessors.

Chapter 5: The Intel Pentium and Pentium Pro

In the chapter, the focus is on the Intel Pentium architecture and start of first original Pentium which was introduced in 1993. It has two integer ALUs and a floating point ALU. It also introduces the level 1 cache and multiple pipelines which shares four stages in common: fetch, decode-1, decode-2 and write. Author then introduces Pentium pro, Pentium II and Pentium III. They were introduced in 1995, 1997, and 1999 respectively. One of key difference between Pentium pro and other processor was location of L2 cache. Pentium pro has L2 cache off die but Pentium II and Pentium III supported caches on die. Overall this chapter provides the good architectural support and comparison between Intel microprocessors and helped me to learn how they actually work.

Chapter 6: Power PC Processors

The focus on this chapter is origin and development of popular family of microprocessor: PowerPC. The 600 series was first in the line and was introduced in 1994. It has cache size of 32KB. It has classics four-stage RISC integer pipeline: fetch, decode/dispatch, execute and write-back. Furthermore it has much simpler front-end than its counterpart x86 at that time. The PowerPC 700 series also known as G3 was introduced 1997. It has cache sizes of 64KB split L1 and dedicated 1MB L2, which leads to its great success. It eventually got widespread into embedded devices and across Apple's entire product line. Then PowerPC 7400 also known as G4 introduce in 1999. It has cache of 64KB split L1 and 2MB L2 supported via on-chip tags. Consequently, it supported compelling multimedia and multiprocessor performance. Overall this chapter has good introduction of PowerPC lines of early processor, their historical development and architecture.

Chapter 7 and 8: Intel Pentium 4 vs. Motorola G4e

Pentium 4 was introduced in 2000, and it was the first major new architecture from Intel since the Pentium Pro while Motorola G4e was introduced in 2001. The two chapters built on last chapters to examine the two microprocessors side by side.

The main difference between two microprocessors technology was pipelining depth which also show the differences of design philosophy and goals of the two processors. In simple term, G4e approach of pipelining was wide and shallow. It has more functional unit in parallel in its back end. Each of G4e execution units has fairly short pipeline, so the instructions take very few cycles to move through and finish executing. While Pentium 4 takes a narrow and deep approach to moving through the instruction stream. Deep pipelining mean that it can hold and work on quite a few instructions at once and it pushes them through its narrower back end at higher rate.

Architectural differences between two processors are also discussed in great length with clearly marked diagrams. Integer units and floating points units explained for two processors. Because of Pentium narrow and deep approach, design is much more suited to floating point applications. One of the key technology is vector execution units on which both G4e and Pentium 4 rely on this for multimedia applications. Overall these two chapters explained the two processors very well and highlighted their differences in a easy to understand manner.

Chapter 9: 64 bit computing

In the chapter, concept of 64 bit computing explained and why we may use them. First of all 64 bit processor means that the processor has general purpose registers that can stores 64-bit numbers. This way ALU and register can handle more possible integer values and more possible addresses. In the nutshell, this will improve the quality of various CAD tools and 3D rendering programs such as weather and scientific simulations, 3D games. But 64-bit comes with a price, that it can squeeze the useful data out of the cache and degrade performance because now the memory address value are twice as large and they take up twice as much cache space.

Chapter 10: IBM's PowerPC 970

This chapter discusses latest line of G5 PowerPC processor which is heart of Apple computers. It has extremely wide back end and 14-stage integer pipeline that built for speed. It instruction architecture is similar to G4e, but it can have a 200 instructions on-chip in various stages on execution as compared with G4e 16 instructions and Pentium 4's 126 instruction. On the whole, this chapter provides the in-depth details of the 970 architecture and provided comparison with Pentium 4 and G4e.

Chapter 11: Understanding Caching and Performance

This chapter introduces a caching in a gentle way. CPU clocks cycles has faster than the memory and bus clock cycles and a great deal of CPU clock cycles has to be wasted in order to retrieve data from the memory. In order to address this problem, cache was introduced as early as first processor came out. Analogy with the warehouse is given as an explanation of cache, and it was a key for me to really understanding the cache concepts. Level 1 cache has an access time of 2-8 ns and consists of SRAM. Similarly, Level 2 cache has an access time of 5-12ns and consists of SRAM as well. When compared with the main memory and hard disk, they have access time of 10-60ns and 3ms-10ms respectively. Cache spatial and temporal locality is discussed in great length. Different Mapping schemes such as fully associative mapping and direct mapping are explained very clearly with simple diagrams. It is very helpful for people who are learning about cache mappings first time.

Chapter 12: Pentium M, Core Duo and Core 2 Duo.

In this chapter, Intel latest high performance and low power consumption processor Core Duo discussed. Chip multiprocessing (CMP) is where two or more processor cores are integrated onto the same silicon die and this commonly called dual-core processor. The focus of these architectures is from single-threaded performance to multithreaded performance. The Intel Core 2 Duo was introduced in 2006 and it has L1 cache of 32KB instruction and 32KB of data. L2 cache is either 2MB or 4MB. It has built in support for 64 bit. The number of improvements has been added to Core Duo processors such as more decoding logic, bigger execution hardware and memory buffer space. Overall, this chapter provided a very good in-depth knowledge of Core 2 processors and it is a must read for anyone interested in the latest development in microprocessors.

Conclusion:

This book is excellent for every level of reader from beginners to engineers with many years of experience and a technical background. Overall, this book had done a very great job explaining microprocessor architectures from beginning to latest Intel Core 2 Duo processors. I wholeheartedly recommend it to anyone who has any interest how computers and microprocessors works.