Chapter 214 Difficulties
The restart of the space shuttle of the FATS program was not smooth, and the same is true for SLS.
After a month of overtime efforts, the new core stage of the SLS rocket has been designed and delivered for manufacturing, but the SLS rocket still has many problems.
This time, its enemy is cost.
As the successor of the space shuttle, the SLS rocket aims to be cheaper than the space shuttle. The original cost of each launch of the space shuttle was about US$500 million. The purpose of the SLS program is to have a larger capacity than the space shuttle, expanding from more than 30 tons to 130 tons, and about US$2 billion per launch, which is more cost-effective and can land on the moon.
Ideals are full, but reality is very skinny.
The price of each of the four RS25 engines used in the SLS core stage is US$75 million, plus the extremely expensive five-stage SRB (solid rocket booster). Claire recently estimated the cost of launching the first SLS at the end of the year. If the cost of research and development and improvement is included, it will never be only US$2 billion.
At least double, more likely more than double, around US$5.5 billion.
US$5.5 billion, just to launch a rocket?
In other words, the three SLS rockets in the early stage of the Artemis program will cost $16.5 billion?
The price of two rockets is more than the entire moon landing budget of China's Shuguang Program!
When Director Claire discovered this fact and the Endeavour renovation budget, he quickly covered up the news and could not let it out.
Only by pretending not to know and waiting until the product is built and cannot be stopped, can Congress be forced to continue to invest in order to maintain the entire project.
NACA has a lot of such experience, so it is not afraid.
…
"EDA software design is also provided by Xinyuan."
On the fifth day of the Semiconductor Summit, July 26, Wang Minjiang once again said the words that made everyone's scalp numb.
The main purpose of the 2016 Semiconductor Summit is only one: to transform the ternary system and surpass tradition.
On the first and second days of the meeting, Huanghe Semiconductor mainly introduced the current ternary technology mastered by Xinyuan.
Except for several semiconductor manufacturers who have already started OEM and speculated, everyone else was shocked that Xinyuan has quietly advanced the ternary technology to such a level.
Chip design, yes. Xinyuan's B-level base automated silicon carbide production line is equipped with dedicated EDA design software, which is very mature and stable.
In order to adapt to traditional silicon wafers and analog field effect transistors, Xinyuan's microelectronics department made a lot of improvements to this type of EDA software at the time. Later, when Huaxin International was outsourced, it used the mask produced by this EDA for lithography.
Basic environment, yes. From SC09, the microelectronics department reversed a set of ternary high-level language systems, which was internally named X language.
X language is different from the current popular languages. It is a mixture of Chinese and English.
From the underlying logic, X language supports both Chinese and English. Most instructions are solved by a certain word, and some complex instructions can be replaced by letter abbreviations. Programmers do not need to have a basic level of English, as long as they can remember some code abbreviations.
If you are used to English code, you can also use it. Since it is based on ternary field effect transistors, for ternary computers, there is no problem that a Chinese character takes up more space than an English letter.
A character and a letter both take up 1 byte in a ternary computer.
At that time, Wang Minjiang also brought two laptops using X32115 chips. They were satellite Internet laptops provided by the company for engineers on business trips. They looked much bulkier than mainstream laptops, but their main functions were only to surf the Internet and modify documents, and their performance was completely sufficient.
In fact, X32115 is comparable to the current mainstream CPU, and can also have graphics computing functions. It's just that there are too few software, and the system is also relatively general, highlighting that it can be used.
The main components of these two laptops, including CPU, memory, and storage particles, are all domestically produced, and they are equipped with the ternary programming software used internally by Xinyuan, which is used for demonstration.
During the demonstration, Wang Minjiang first briefly explained the X language, and then invited several participants to try to compile the program on the spot.
This kind of programming in both Chinese and English made them feel very wonderful. Although the codes written during the experience were just small tools for introductory learning, they could still feel the maturity of the X language.
The high efficiency has not been clearly experienced for the time being, but it is obvious that the X language programming is not complicated, at least compared with the mainstream C series language and Python, it is not bad, it is easy to get started, and it is quick to learn.
On the contrary, due to the closed-mindedness of its founder, Easy Language has been out of date for many years, and even does not support the X64 architecture. Who would still use that old thing?
Although the simplified operating system removed by Xinyuan from SC09 has average performance, it is also smooth and has no bugs. It just lacks a software ecosystem, which is also very surprising.
After five days, the confidence of the companies participating in the summit has gradually grown from nothing to something bigger and bigger, and even started to swell.
Originally, they thought that ternary systems were very unreliable, but Xinyuan had already mastered the basic process of chip design, underlying logic, operating system, and program compilation, which was just relatively monotonous.
Although their most attractive XW series equipped with AI system did not reveal any information, it was enough to let people know the potential inside.
For other difficulties, Huanghe Semiconductor, which has been authorized by Xinyuan with a large number of patents, only said: "We have it."
This is not a technical breakthrough, it is a platform waiting for them to perform.
After a few days of the summit, Shanghai Huahong said that it will focus on ternary memory and storage particle manufacturing, Tsinghua Unigroup and Zhongke Loongson decided to develop desktop and mobile CPUs with ternary architecture, and Jiangcheng Digital Engineering Research Institute planned to immediately study graphics processors and AI processors with ternary architecture.
Huaxin International... Huaxin International looked at Shanghai Microelectronics, or everyone present looked at Shanghai Microelectronics.
He Yuming, the general manager of Shanghai Microelectronics who came to the meeting, was a little embarrassed, because if this summit was a shot of encouragement to integrated circuit manufacturing and design companies, it was a death warrant for Shanghai Microelectronics.
After five consecutive days of rigorous discussion, everyone came to a unanimous conclusion that even if the advantages of the ternary system were fully utilized, the difficulties of backward process could only be avoided in a small range.
Now 28nm ternary constraint is equal to 14nm binary. If the world mainstream enters the era of less than 10nm, the ternary system must at least enter 20nm or 18nm.
That is to say, it is now 2016, and by 2020, the minimum processing process of domestic lithography machines must reach within 20nm.
If not, then the transition to ternary system is a failure, and it still cannot reverse the backward situation.
The current 90nm lithography machine is not good, and the upcoming 65nm lithography machine is not good either. What is needed is a 40nm or 40nm++ lithography machine to barely reach the 14nm process.
However, can He Yuming guarantee that a 40nm lithography machine will be delivered before 2020?
Not only the lithography machine, but also the requirements for chemical raw materials have greatly increased when entering the process within 20nm. Can China keep up?