Chapter 255 The most high-end technology often only needs the simplest solution
Monday, October 8th, the first working day in October.
Countless migrant workers dragged their ‘tired’ bodies into the company and suffered from post-holiday work syndrome overnight.
They all looked listless.
Classmate Wen Liang first sent Li Rang to the primary school attached to China Normal University, and then went to school.
It was a bit early, and there was no one in the classroom. If Wanmi hadn't been worried about comparing the class schedule with the classroom, he would have suspected that he had gone wrong.
After arriving, Wen Liang didn't bother anymore. He sat in the last seat in the last row as usual, took out his mobile phone and scrolled through the range.
During the National Day random travels, Wen Liang and Su Jian often turned into evil spirits and stirred up trouble in the range.
Not a word was mentioned about the previous insertion.
Suffocated a bunch of people to death.
Mainly, the user interface optimization of the most superficial level of the Star system is still done bit by bit, without any major breakthroughs, and it is unlikely to disgust Apple.
What the two of them did was illogical.
Making trouble everywhere and stirring up dissension.
Poland, which had not joined before, also came to the range, and finally the five permanent members of the range were formed.
In fact, the development of range is not very smooth, and the Tour America-style development seems to be a little less stylish in the eyes of overseas users.
Moreover, Wen Liang has a rhythmic plan for the development of range. The core of the current stage is to keep an eye on major overseas manufacturers. Once a major manufacturer stops making short videos, he will immediately start fighting to increase the popularity.
So much so that range has not been launched yet. After many adjustments, we have found the recommendation algorithm that can best capture users.
Just as Wen Liang was teasing Ah San, whose fighting prowess was off the charts, for a while in the range, seven or eight classmates swarmed into the classroom.
The first person looked up and saw Wen Liang, and was a little surprised: "Hey, Wen Liang came so early today."
"Mani Mani, why don't you go in? Wen Liang came early enough today."
"..."
The chatter quickly became lively.
After a while, Chen Jiaxin, Tang Pu, Zhao Mingjie, the eldest sister Liang Yameng and others also came.
The topic quickly turned to the 'star system'.
Although the activity decreased during the holidays, one version per day was used to optimize the user experience, which indeed gave users surprises.
"Wen Liang, there have been speculations on the Internet that the 'Star System' is related to Bolang. Is it true? I asked someone to help me refresh the Star System. In addition to frequent updates, the usability has indeed improved."
"Bo Lang doesn't usually work overtime, and it's even less likely to work overtime during the National Day holiday. I don't think it's the case. Many big Vs on the Internet have denied their relationship with Bolang because of this."
"Last semester, there were rumors in the school that Bolang was working on solving some technical problems in the laboratory cooperating with the school. Could it be the Star System?"
"No, it's just an Android UI. It doesn't need to be treated with such high standards. Besides, it's clearly engaged in integrated circuit design."
"But there is no place where Bolang needs to make IC."
"Then who can say for sure?"
One of the students put forward opinions, and another expressed objections. Wen Liang said with a smile on his face: "The Star System is related to Bolang. It's hard to tell everyone the more specific details."
Everyone was surprised and happy.
"It turns out it really does matter!"
"No wonder it becomes more and more useful. I know that only Bolang in China has the ability to adjust the most outstanding UI."
"Well, Bolang can make mobile phones. If it is just a UI and not put on the mobile phone, it will be a bit rootless."
"Definitely, that's why Bolang joined forces with the school to conduct IC design research!"
"But didn't Qin Zheng and Academician Cao go abroad to study?"
No one wanted to get any answers from Wen Liang, and Wen Liang didn't say anything more.
…………
…………
In the afternoon, Wen Liang walked into the company.
The office atmosphere is as usual, everyone is very busy.
Sitting behind the desk, Wen Liang flipped through several documents on the desk.
Everything expected was in place.
The progress of the IDE (Integrated Development Environment) supporting the Star System is acceptable. The first beta version has been completed. After completing all aspects of testing, the official version can be launched.
The second phase of technical optimization of the Star System has begun.
However, the graphical interface of ‘World’ has been taking shape for a long time. The plan is to eventually form a system from point to point, so the marketing efforts will be slightly slower.
Li Ze submitted an additional draft.
He believes that once again vigorously promoting the system should be coordinated with the promotion of IDE and endorsed in the name of Bolang.
It means that this requires a suitable timing, which will come from some actions of Alibaba Cloud OS.
There is nothing wrong with the direction of this draft.
Nowadays, the promotion of the Star System has reached a bottleneck. To break through, there really needs to be a big movement to endorse it.
Temporarily shelving promotion is also an option, and the scheduled IDE promotion meeting was postponed as a result.
After opening the last report document and taking a quick look at it, Wen Liang immediately sat up straight and shouted directly outside: "Secretary Wan, inform Mr. Ze, Mr. Tang, Mr. Chen, Mr. Wu, Mr. Zheng, and Mr. Xiao to come immediately.
my office."
Wan An immediately informed him.
Several people arrived at Wen Liang's office in minutes.
Seeing everyone, Wen Liang got straight to the point: "Have you reviewed this plan?"
"The process has been approved." Chen Jiaxin answered first, "Did something go wrong?"
Wen Liang frowned slightly and asked directly: "Are you sure this is the SoC tape-out including the CPU on our company's own mobile phone, and not the tape-out of other chips on the phone?"
Next to him, Li Zeqi said: "Isn't this already in your plan, Mr. Wen? It's also a request you made before the holiday."
Wen Liang was still a little confused: "Wu Gong, can you tell me how this trick works?"
Wu Gong is Qin Zheng's deputy. Recently, he often works with the hardware development department of Bolang Lab's mobile phone business department in the company.
"Mr. Wen, the SoC chip to be taped out this time is not self-developed." Wu Gong understood clearly. He knew that there was a slight misunderstanding in the description of the plan.
"In the past three months, a large number of people in Huagong IC Laboratory have been doing research on the ARM v8 CPU architecture and mali GPU architecture, and have completed a lot of design work. After Qin Gong went on a study tour last month, everyone was full of energy.
Put it on this, experimental results were quickly produced, and the company completed the integrated design and verification."
"You mentioned before the holiday that necessary chips cost money, so we worked overtime during the holidays, conducted multiple rounds of verification, and finally approved and proposed a tape-out plan..."
As Wu Gong explained, and Zheng Gong and Xiao Gong added from the side, Wen Liang understood the whole story.
It’s an ‘own goal’.
………
In this era of 2012, the global mobile market has actually just begun to rise.
Even if Apple’s mobile phone is in its fifth generation.
It is still an era where there are many players and they are divided against each other.
The former mobile phone empires collapsed one after another, and new players sprouted up like mushrooms after a rain. Except for Apple, the rest did not have much success.
There are many players in each core field.
In terms of systems.
iOS leads the way.
Nokia, who did not believe in evil, tossed through multiple systems and finally chose to embrace Microsoft's WP.
Samsung took over MeeGo, which Nokia gave up, and then integrated Tizen with Intel using two major Linux alliances and Samsung's self-developed BADA. It has been unable to compete with iOS and Android, but it has never given up.
The BlackBerry mobile phone, which is highly supported by America's top management, and the BlackBerry system are now in full swing in South America.
Google took the lead in forming the Open Handset Alliance with hundreds of hardware, software, and communications manufacturers, and established the Android Open Source Project (AOSP) based on Android to promote Android globally.
Later, it was regarded as a standard by domestic players who like to advocate overtaking on corners.
Some manufacturers, led by Lenovo, have developed so-called self-developed mobile systems based on this, and have received support from core high-end infrastructure projects.
Including Alibaba Cloud OS, which is currently also known as a self-developed system.
Boss Lei, who was considered by later generations of netizens to have no big picture, saw that it was very troublesome to develop a mobile phone system by himself, so he got inspiration from Meizu Huang Zhang, and finally built MIUI based on Android, and now it has the advantage of being the leading domestic UI.
The convenience of being able to use native Android directly has led some other low-end domestic mobile phone manufacturers to place bets.
As a result, Google’s Android has gained 90% market share in China.
In terms of core processing chip, SoC.
Apple has expanded its own A-series chips based on the ARM instruction set architecture, which is also the leader.
Among the flagship models, the Samsung S3 is based on the ARM architecture and designed and developed the Exynos (Orion) chip, and the HTC One X uses Nvidia's Tegra3.
Other models on the domestic market, namely the Mi 2, have reached flagship standards and use Qualcomm chips designed and developed based on the ARM architecture.
In addition, there are Texas Instruments, MediaTek and Juchang who are polishing K3V2.
Based on design and R&D, there are currently about ten players in the world who can be ranked.
but!
But here comes the very interesting part. Taking the domestic market as an example, Google Android occupies 90% of the market share, right? In fact, the only flagship phone available is Xiaogaoliang's Mi 2, which has not been officially launched yet.
Meizu has half of the MX model because there are too many revisions.
Even counting the entire year of 2012, the two companies combined probably only had shipments in the tens of millions.
Therefore, except for Apple, which is relatively awesome because it has subverted the mobile phone industry, the rest are all the fault of the elder brother, don’t laugh at the second brother.
In short, when all the information was integrated and presented to Wen Liang, he found that he had many deviations in his judgment of the actual market in 2012.
I won’t talk about the system layer.
Once the Star system was launched, it directly aimed at flashing Android series flagship models, and has covered more than two million machines, which is enough to show that 90% of Google Android's market share in China has been covered in important markets.
As for processing chips, if we first solve usability problems, the research and development costs will be affordable, and the research and development cycle will be really short!
The current situation of ARM, which is famous in later generations, is not very impressive.
In the first three quarters of 2012, ARM’s sales were only US$500 million. Note, these are sales!
The current market value of ARM fluctuates around US$12 billion, and its assets fluctuate around US$1.9 billion.
As early as July, because the IC design led by Qin Zheng was in a deadlock, Wen Liang approved the purchase of ARM's IP license.
License fee of $8.7 million and royalties of 1.5%.
According to ARM’s financial report, revenue comes from 33% licensing fees, 50% royalties, and tool and technical support.
At present, licensing fees generally range from one million to tens of millions of dollars.
This price is basically top-notch.
At this stage, ARM's IP authorization for core processors is divided into three forms: architecture/instruction set level authorization, core (IP core) level authorization, and usage level authorization.
Architecture-level authorization means that the ARM architecture can be significantly modified, and the ARM instruction set can even be expanded or reduced.
Apple is a typical example, having expanded its own swift architecture based on the Armv7 instruction set architecture.
Core-level licensing means that you can use cores such as Cortex-A53 as a basis and then add some of your own designs to finally form an SoC. Qualcomm and Juchi are typical examples.
The difference in authorization at this level can be huge.
For example, Qualcomm has its own GPU architecture, so it does not need GPU kernel authorization, and it even has the idea of magically changing the CPU architecture.
Juchang does not have a GPU architecture and needs authorization in this regard. Of course, this authorization can be purchased from ARM or other companies.
After all, SoC is a comprehensive layout.
The final usage level authorization means that you cannot change ARM's IP, nor can you use ARM's IP to create your own packaged products based on this IP.
A simple metaphor:
An article A can be modified and used, which is an architecture-level authorization. B can only quote and use it, which is a kernel-level authorization. C can only forward but not change, which is a usage-level authorization.
The consideration at that time was that it was suitable for Qin Zheng and others to use as a practical reference for research, so Bolang purchased a complete set of architecture and kernel-level licensing content.
That is, the Armv8-A CPU architecture and mali GPU architecture launched in November last year.
Including IP cores such as v8-A57, v8-A53, v8-A35 and mali-400.
This is also the reason why licensing fees are so high these days.
This part of the copyright purchase fee is counted as the cost of research reference materials in basic research funds.
Of the 350 million basic research funds as of the end of July, the IC Laboratory has spent 270 million, including the US$8.7 million in licensing fees totaling 55 million yuan.
Qin Zhenghui's previous rejection of the visiting scholar invited by Academician Cao was related to this.
Because there is a huge gap between the ultra-high investment in basic research data, research environment, research equipment and other aspects of chips and the constant overthrow and restart of research into a deadlock.
Later, after the launch of the Star System, everyone became passionate, and with gentle persuasion, Qin Zheng finally ‘awakened’.
Perhaps the goal of investing a lot of R&D funds from the beginning was to conduct self-research from scratch, so Wen Liang subconsciously did not consider the option of developing SoC based on ARM-licensed designs.
Also, Wen Liang misjudged the market situation in 2012. If we designed and developed a chip based on ARM's in-depth authorization, the cost of trial and error would not be high.
Take the current full set of in-depth authorizations that Bolang purchased at a large price as an example. Three months is not a short period of time. The fastest is to buy it on the same day and lose it to TSMC Current Chips.
It is 100% guaranteed that the tape-out will be successful. As for whether it will be useful or not, no one can guarantee it. There are many examples of overturning.
Now Huagong IC Laboratory and Bolang Laboratory have spent more than three months and a lot of research and development funds, and the preliminary approval results should be quite good.
After all, the main research content of Huagong IC Laboratory in the past three months is to try to use these IP cores for permutation, combination and small-scale tuning.
A more direct description: This method of researching content is the simplest exhaustive method.
There is no shortage of research and development funds, but there is a lack of time, so each team tries separately, arranging and combining them according to 1, 2, 3, and 4 to see which one is easier to use.
It should be noted that this SoC to be taped out has very, very few things designed by Bolang.
In short, if you just want usability, RMB 100 million can roughly buy an SoC using 32nm or 28nm technology - in fact, it can be seen from the process that the investment will not be that high.
However, in Wen Liang's view, as of the third quarter, the two laboratories had spent as much as 460 million on basic research, 75% of which was spent in the IC laboratory. He also thought that the investment would be very high based on previous concepts.
I think it will cost at least several hundred million dollars.
Who would have thought that chip design, a technology considered to be very high-end, can be solved with a very rudimentary and violent exhaustive method, whether it has IP authorization or not.
It’s no wonder that the laboratory burned nearly 350 million of what could have been burned with 100 million yuan originally, and is still in the process of tearing it down and starting over.
So, Wen Liang was so rude when he came to the company today and saw the SoC tape-out plan.
After clarifying the ins and outs, Wen Liang did not hesitate and said directly: "If there is no problem with the technical review, we will take the TSMC current chip. If the tape-out fails, we will continue until it succeeds."
"Compared with the current expenditure, if our first mobile phone can have higher control over the core hardware and software, then we will not be far from business success."
After talking about the arrangements, Wen Liang directly kicked everyone out of the office without giving everyone a chance to say more.
After closing the door, Wen Liang rubbed his hands and said excitedly: "Haha, I turned out to have read the wrong era! It's over for those who step on horses!"
-
PS: Regarding the exhaustive method, this is also done in some higher-end projects, because even in the technological development of 2022, there is actually very little innovation, and many key projects are solved by violence, including materials, hardware, and software.
For example, recommendation algorithms are generally violently built using massive amounts of data.