I recommend a book to students interested in processor design

First slice

I have always thought that this book should be called "dedicated processor-centric SOC design", because it does not mean "complex SoC design", but it also means a literal translation of the English name, maybe the author thinks his SOC design philosophy is relatively complicated, or it is specially designed for complicated applications. Let's talk about the source of this book first.

In retrospect, I was still a graduate student four years ago. I didn't know why. I borrowed this book from the library and read it at hand. Perhaps because of Chris Rowen's speech in Shanghai, he specifically mentioned the similarities and differences between tensilica's design philosophy and FPGA, which gave me a strong curiosity when I blindly worship FPGA, unfortunately, the design of dedicated processors was too simple to understand what the book was about.

The book is often translated, and I know some terms. Although I have been influenced by the SOC design philosophy of tensilica, there is actually no practical experience. I thought it was just a matter of thumb. Later I realized that it was just a matter of novelty. However, this book, the company, and the author have an inexplicable yearning.

After graduation, I borrowed this book from my organization. I don't know why, but I always think that the book has its own "Yan ruyu ". It wasn't until you started to get into touch with the processor design project that was just as open-minded.

Let me start with the classic "Computer System Architecture: Quantitative Analysis Method" (everyone on the Earth knows that this book is a strange book in the computer industry and has been released for four editions in the past thirty years, first, I was shocked by the president of Stanford University (one of the authors of the book, and the other author is the inventor of the book.) for the first time, I felt the high threshold for processor design, involving a wide range of technical fields, this is not enough. According to my understanding, the processor design engineers must constantly make multiple choice questions and choose iterative multiple choice questions. Once a wrong choice occurs, the entire design will collapse. My generation can only look up to the sky to sigh.

Let's take a look at what is a dedicated processor design. A dedicated processor is also called ASIP. Typical examples include GPU and NP. This concept is easy to understand. It designs a set of instruction sets (corresponding to a processor architecture) for a certain type of applications ). Unfortunately, if you really want to get a processor from a specific application, it's so difficult (why is the application king? Is it ...), to learn more about this, read the two books processor design SOC computing for ASICs and FPGAs and embedde DSP processor design.

If you can read the two books smoothly, you will be shocked by the odd book. Read complex SoC design again. At this time, I believe you will understand most of the content. Of course, you will again sigh with Dr. Chris Rowen, who is still insightful. To put it bluntly, he is going to drop the design threshold of a dedicated processor and reduce the design cycle by one (from the book you can understand that designing a processor is the most complicated ). He cleverly chose a processor prototype/Instruction Set (xtensa, also said to be from MIPs). You can repair and complete the prototype for your target application, in particular, they provide a powerful software tool. You only need to describe your processor in one advanced language, and then automatically generate the tool chain related to the processor and the final RTLCode/Network table. The entire process is much easier than starting a new instruction set/architecture.

Back to this book, Chris Rowen talked about his views on SOC design in the first three chapters. Chapter 4 talked about the challenges of complex SoC design. In fact, it is mainly about the design of multi-core processors, hardware and software design and implementation of communication between processors. Chapter 5 describes tensilica's dedicated processor design from the perspective of hardware and software, Chapter 7 introduces the design of the processor and multi-core, and describes the advanced languages they use to describe the processor. Chapter 8 looks forward to the future SOC design, and puts forward that he uses Professor Christenson's disruptive innovation theory to analyze their design philosophy. To be honest, I really admire this old boy. You'll have to sigh again when you read his book about disruptive innovation. Whining, my generation...

One sentence in this book (many people have already said): It is not a simple design process book that embodies a SoC design philosophy. In fact, it is necessary to give you a processor design process.

I think reading technical books is the same thing: a person, he does not understand any more in-depth academic works. He only reads Dan's books and only Yi Zhongtian's books. This is neither a mistake nor an immoral one, this is probably only unfortunate. The more you read, the more you find that you really want to read the so-called classic classics in our minds. You may have a little luck. For example, if you want to grow up in a good family, have a good family education, and have a good education in both primary and secondary schools, and the premise is that you have a chance to get an education. You slowly walk through this lucky track and cultivate a kind of reading ability through the development of education, which helps you read books that many people cannot read.
(it is strongly recommended that you check the Chinese version in English, you know why !)