"Well, since you're so anxious, the surprise I'm talking about is really not referring to a computer workstation, but a new set of instructions!" Li Xuan said.

"God, is this true?" said John Hennessy, his eyes widened and he looked at Li Xuan in disbelief.

The young man in front of me just developed the ERM instruction set last year, and he is working with himself on the MIPS instruction set. It's unbelievable that he's still working on a third new instruction set. Hennessy stubbornly used his own self-cultivation to suppress the "**" that was about to say it.

"Haha!" Li Xuan looked at the Stanford principal in front of him, his face was like constipation, and he couldn't help laughing, "I'm kidding you, it's not really a new instruction set! How can it be so easy to develop a new architecture, I am now focusing all my energy on the research and development of the follow-up architecture of ERM."

The so-called new instruction set is actually a new subset of 16 bits of instruction length based on the original 32-bit instruction set when my team developed ERM-2. I'm going to name this new subset thumb. Compared with the equivalent 32-bit instructions, the thumb instruction set retains the advantages of 32-bit instructions while greatly saving the storage space of the system. ”

"A 16-digit instruction?" Hennessy pondered.

Yes, the operand is still 32 bits, and the command address is also 32 bits, but some of the features of the original ERM instructions are discarded. However, the thumb instruction set does not change the underlying programming model of the ERM system, but only adds some restrictions to the model, as long as certain invocation rules are followed. The thumb subroutine and the arm subroutine can be called to each other. Li Xuan explained.

"That's a great idea!" Hennessy, a tech guru, naturally understands a little bit. "I'd like to dig into your thumb directives, if possible in the subsequent development of the MIPS architecture. It is also possible to implement a 16-bit 32-bit mixed instruction set. ”

"Our previous tests have shown that the processor only needs 60% of the storage space required in the ERM instruction mode when entering the thumb instruction mode, and the storage rate of the former is 50% faster when using 16-bit memory!" Li Xuan continued.

In the past, 64K and 128K refer to the storage capacity of the memory chip, which determines how much information the memory can store. The 16-bit and 32-bit refers to the bit width of the memory, which determines the speed at which the memory chip stores information. Generally speaking, chips with larger bit width and larger capacity are more expensive.

For memory chips of the same capacity, the larger the bit width. The higher the price. The mainstream memory chips on the market are all 16-bit. The ERM architecture is 32-bit, which means that the output speed of the ERM-1 processor is faster than the speed of memory reception, which creates a storage bottleneck and affects the performance of the ERM processor. In general, the performance of the computer can only be perfectly utilized when the bit widths of the processor and memory are the same.

"It's a pity that Thumb is not a complete instruction set after all, and he doesn't use it very frequently in the entire ERM instruction set!" Li Xuan shook his head and said.

The 16-bit thumb instruction is just one of the means that Li Xuan solves the memory bottleneck that 32-bit processors are currently encountering. He has brought up the concept of caching. Subsequent ERM chip design will add cache to alleviate the contradiction that the memory rate cannot keep up with the processor rate. Of course, space inside the processor is extremely valuable, and the cache is not integrated into the processor today, but comes in the form of an off-chip cache.

Hennessy can't wait to learn more about the thumb instruction set. So Li Xuan found him detailed relevant information, and the two spent an afternoon in the discussion to satisfy Hennessy's curiosity about the new technology.

"This is one of the best Chinese restaurants in Hong Kong!" Li Xuan specially booked a sumptuous dinner table at Fu Lin Mun Restaurant for Hennessy to wash the dust. But the foreigner in front of him has little interest in the table full of Chinese food.

Li Xuan couldn't help but shake his head, obviously the other party was not a lover of Chinese food. Not every foreigner will give a thumbs up to Chinese food. The two exchanged while eating, and Li Xuan invited the other party from Silicon Valley to Hong Kong. It's not just to show off the new breakthrough in the development of the ERM-2 architecture.

"John, I want to invite you to come to Hong Kong to work!" Li Xuan directly stated his purpose.

Hennessy looked at Li Xuan, shook his head and said, "I may not be able to adapt to life in the East. For example, I don't like the food here, and I prefer American red wine steak!"

"There are a lot of Western restaurants in Hong Kong, if you are willing to try Chinese, French, Japanese, Korean, Thai, and Indian food, the diet is not a problem!" Li Xuan continued, "Stanford University and Hong Kong Chinese University are going to carry out an exchange program, and they will send teachers and students to visit each other every year in the future!"

You can use the professorship of Stanford University to teach at the Chinese University of Hong Kong for a few years, and at the same time continue to carry out the follow-up development of the MIPS project at the Hong Kong R&D Center of the Institute of Oriental Studies! I think there are many things that the ERM team and the MIPS team can communicate and discuss with each other. For example, the thumb instruction set I just mentioned. Another example is the special divider design in the MIPS architecture, which can execute division instructions!"

After the first S-100 processor based on the MIPS architecture was finalized, the cooperation agreement between Stanford University and the Oriental Institute has ended. Li Xuan began poaching Hennessy's R&D team as early as last year.

More than half of the team members will resign from Stanford or join the Oriental Institute's U.S. R&D Center after graduation. As the core person in charge of the entire project, Hennessy is naturally the focus of Li Xuan's research. He offered the other party a very good treatment, but Hennessy did not give a clear answer.

"If it's just a short-term exchange, that's a good idea!" Hennessy said after a moment's thought. He also felt that if he could have an in-depth conversation with the ERM development team, he would definitely get a lot of useful inspiration.

The ERM-1 processor developed based on the ERM architecture has been a great success on the ABC series of computers, and the Hong Kong ERM development team has begun to make a name for itself in the entire IT field. It's not just Li Xuan and the Oriental Research Institute that have been wiped out by the profits. Like the Chinese University of Hong Kong, it has also received a lot of benefits.

In the early stage of the development of the ERM project, Li Xuan did not directly come to the forefront. Professor Chiu Wai-ming, then Associate Head of the Department of Electronics at CUHK, was responsible for the research and development of the project. The mailing address of many papers is the beginning of the Chinese University of Hong Kong. Later, after Li Xuan personally took over the ERM project team, Zhao Weiming was still one of the co-leaders of the team.

Hence the Department of Electronics at CUHK. It has also gained a lot of fame through the ERM architecture. CUHK's recent academic exchange agreement with Stanford University focuses on the field of electronics. Although Dongfang Electronics has done a lot of work behind the scenes, the strength of the Department of Electronics at CUHK can be recognized by Stanford University, which is the premise for the exchange and cooperation between the two sides.

"ERM and MIPS are the world's first and second RISC (Reduced Instruction Set) processors. Have you ever considered writing a professional book on computer system architecture based on the research and development of these two processors, combined with the ABC series of personal computers of Aikang Company, and the upcoming workstation computers?" Li Xuan suddenly asked.

A gleam flashed in Hennessy's eyes, and it should be every professor's dream to write a book. Hennessy was an absolute genius, earning a PhD in electrical engineering at the age of 25. He then joined the faculty of Stanford University and became a tenured professor at Stanford University at the age of 28. Now he has just turned 30 years old, and with the strong support of the Oriental Research Institute, he has led a first-class scientific research team to successfully develop a new RISC architecture processor.

Now Hennessy can only be regarded as a calf in the academic world, and the temptation to write his own book is naturally not small. But to write an excellent book, it is by no means an easy task, at least the accumulation of personal knowledge is very important.

in another time and space. John Hennessy resigned from Stanford in 1984 to found MIPS Technologies, a company that commercialized the MIPS architecture and developed a series of processor products. Advanced workstation computers for companies such as SGI.

He then returned to Stanford University as the director of the Computer Systems Office. After achieving both business and academic fame, he joined Professor Patterson, another computer bull at the University of Berkeley. He co-authored two textbooks, which became compulsory courses for all computer science students in the United States.

But now Hennessy has not yet become a big bull. Not to mention President Xu of Stanford, it is not easy to write an influential textbook on his own. If you follow the gimmicks of ERM and MIPS, the only RISC architecture in the world. Coupled with the use of the world's best-selling ABC series computers as an example of textbook analysis, the impact of the new book will be much greater.

The reason why Li Xuan encouraged the other party to write a book was because it was also a good thing for Dongfang Electronics Company. If the quality of the new book is good enough to be selected as a textbook by American universities, then the influence of the ERM architecture, MIPS architecture, and ABC series computers will spread to the entire computer industry in the near future.

When computer science students who use these materials graduate, their familiarity with both architectures is much better than the others, which translates into an implicit advantage for ERMs and MIPS to compete with other RISC processors in the future.

"Are you going to write a book with me?" asked Hennessy earnestly, he didn't think that one person could do the whole book, and the man in front of him was a good person to work with.

This temptation is also not small for Li Xuan, he has made enough money. It would be exciting to think that he could write a textbook that would be a must-read for all computer students in the United States, and that his prestige would be remembered by every student for decades to come. He can even translate books into Chinese, Japanese, French, German and other languages, spreading his influence to the whole world.

"It's a great idea, let's just write an excellent textbook that every computer student must read!" Li Xuan said loudly.

At this time, Hennessy finally decided to come to Hong Kong to work for a while. The reason why Li Xuan tried his best to invite the other party was, in the final analysis, because of the lack of talents in Hong Kong. The square-faced white man with thick eyebrows and big eyes in front of him is an all-round talent with excellent IQ and good emotional intelligence.

In another time and space, Hennessy not only had excellent technical prowess, but also ran a good company, and later successfully became the president of Stanford University. It is not easy to be a university president in the United States, and in addition to the complicated daily affairs, another important job is to solicit sponsors and donations. Presidents have to keep communicating with all sectors of society, especially the government and alumni, and they are simply not competent if they do not have enough skills and skills.

With such a comprehensive talent, Li Xuan naturally has no reason to let go. Professor Zhao Weiming, the vice president in charge of academic research and development of the Institute of Oriental Studies, has not kept up with the speed of development of the Institute of Oriental Studies. This is one of the reasons why Li Xuan personally took over the project in the later stage of ERM architecture development.

Professor Zhao Weiming is slowly transitioning to the administrative side and is no longer involved in specific research projects. The Oriental Research Institute needs a new academic leader in processor research and development, but Li Xuan himself cannot put all his energy into research and development, after all, he still has a huge Dongfang Electronics company to manage.

John Hennessy is Li Xuan's search for the academic director of the Oriental Institute for the next few years. The next two or three years can be described as the most critical years for the development of ERM architecture and MIPS architecture. If the two structures can establish a dominant position in the overall market before other companies can launch other great RISC architectures, it will be much more difficult for other companies to challenge them. (To be continued......)