my return to life

A simple opening remark, and then introduced the identities of Fang Nian and others.

Except for a few insiders, most people knew that Fang Nian was Fang Nian for the first time.

Although Shang Nian was present at the unveiling ceremony just now, there was no introduction of Fang Nian.

At the end, Shen Demin briefly reported on the administrative work of Bai Ze Laboratory: "Mr. Fang, the administrative work of Bai Ze Laboratory is roughly like this."

In fact, there is nothing to say.

Shen Demin is very clear about what is the core of Baize Laboratory, and the administrative team is a coordinating role.

Moreover, Bai Ze Laboratory does not have independent financial rights.

Although the administrative team includes human resources and financial teams, it is only responsible for the financial and accounting work of laboratory employees and does not have independent financial distribution rights.

To sum it up in four words:

Cooperate well.

Fang Nian nodded slightly, but did not speak.

Shen Demin started the second process clearly: "Professor Zhang, please make a summary report on the progress of the research and development work in the laboratory."

Professor Zhang did his part.

Opened the folder, and made a report one by one: "Mr. Fang, Mr. Lu, and Mr. Guan, the research and development work in the laboratory is progressing smoothly;

……

Through the unremitting efforts of the team, the entire design process of the smartphone SoC has been completed. "

"..."

After Professor Zhang finished his report, the conference room fell into a moment of silence.

Fang Nian waited for a while and found that there was no more to say, so he couldn't help frowning: "It's just gone?!"

Then he said in a more emphatic tone: "What about the finished chip design diagram? What about the expected parameters and performance indicators of the chip design? What about the tape-out plan?"

"I don't understand the internal design of the chip, but I can still understand the functional modules, parameters, and performance indicators to be achieved?!"

After Fang Nian's words fell, the meeting room suddenly heard a pin drop.

Everyone felt the pressure from Fang Nian.

Professor Zhang, who was over half a century old, got up quickly: "Sorry, Mr. Fang, please wait a moment, I forgot."

Academician Chen, who is also over fifty years old, also hurriedly said: "I'm really sorry, we are not prepared enough."

"..."

Fang Nian didn't say much, just wait for the next article.

The atmosphere in the conference room suddenly became tense.

Shen Demin is quite strange, shouldn't he be prepared?

Fang Nian has also browsed the laboratory summary report, which is the specific progress of each stage, so there is no problem.

But he really didn't expect this to happen at the meeting.

Soon, Professor Zhang opened the USB flash drive he carried with him through the conference room computer.

This is data with low confidentiality, but there is no too strict standard, and it should have appeared in the meeting, so there is nothing wrong with it.

But Professor Zhang still explained: "This is information that can be disclosed in internal meetings, and it was brought out in strict compliance with the confidentiality regulations."

"In principle, the finished chip design drawings are not allowed to be brought into the conference room, so there are none here."

Fang Nian nodded, indicating to continue.

Professor Zhang quickly started the PPT presentation.

"This chip, internally named Baize #01, is designed according to the 28nm process standard...

CPU adopts ARMv7 instruction set, cortex-A9 architecture...

Integrated PowerVR SGX540 GPU…

Integrated baseband WCDMA, TD-SCDMA, CMDA2000..."

"..."

"The main frequency of the CPU is 1.3Ghz, 4 cores, and the processing performance is expected to be 5% of the Apple A120 chip..."

"Supports 32-bit dual-channel LPDDR2 800mhz memory..."

"..."

"Supports 800-megapixel camera, supports 1080P display, supports NFC, GPS, Bluetooth, WiFi..."

"According to the established goals of the project, the performance of the graphics processing performance, processing performance and other aspects of the design scheme will comprehensively exceed that of the Apple A5 chip by 20%."

"..."

From this point of view, Professor Zhang really just forgot for a while.

After all, the PPT is well prepared.

After everything was said, Professor Zhang finally showed a mobile phone SoC simulation diagram through projection, showing various components.

into two columns.

From top to bottom on the left side are: baseband namely: modem and Wi-Fi, digital signal processor (DSP), audio codec, system memory (RAM);

From top to bottom on the right side are: graphics processing unit (GPU), central processing unit (CPU), image signal processing (ISP), multimedia and navigation positioning module.

As can be seen from this rendering simulation, the mobile phone SoC is a complex integration solution.

Almost a ratio of one to one, the CPU only accounts for 15% of the entire SoC.

The rest are all other functional modules.

From these parts alone, Bai Ze Laboratory really has a long way to go.

Looking at the logos of different companies on this picture, Fang Nian remained silent for a long time.

The meeting room fell into silence.

After a long time, Fang Nian smiled and said, "Does the proprietary intellectual property account for 5% of this?"

Facing Fang Nian's gaze, Academician Chen coughed twice: "At present, it is roughly estimated that it is less than 10%, and it must be a little more than 5%."

Fang Nian laughed twice, and said in a joking tone: "Suppose, I mean, if starting tomorrow, all chip solution suppliers including arm company refuse to cooperate with Bai Ze, how long will it take for Bai Ze to do so?" Make a decent chip."

"According to the current technical reserves and accumulation, in extreme cases, it will take at least one year." Academician Chen also replied with a smile.

"It may be faster, because we have not fully absorbed the technology accumulated by Bai Ze Lab. After all...the chip numbered 01 now has a very strong requirement for the mass production time of the finished product."

Then Academician Chen explained: "To put it more harshly, the laboratory is only an integrator this time;

This is what almost all chip design companies do, based on the infrastructure of each solution provider for their own design;

After all, it is now a global collaboration. In the case of foreign solution providers such as Qualcomm and arm, there is no need for commercial companies to invest repeatedly. "

After hearing Academician Chen finished speaking, Fang Nian nodded with a smile: "That's true, after all, it's too wasteful to reinvent the wheel."

Having said that, Fang Nian sighed inwardly.

In fact, only when we are really involved in this industry can we know how high the overall knowledge barrier is.

It is too difficult for a commercial company to do everything again.

It is not difficult to understand that China has not produced a semiconductor company with completely independent intellectual property rights from the beginning to the end.

Including the HiSilicon Kirin that attracted much attention later.

The CPU architecture is arm, the instruction set is arm, the GPU is the mali series acquired by arm in 06... and so on.

To put it bluntly, outside of Qualcomm, that is, Apple's self-reliance is relatively high. After Qualcomm, it uses its own GPU.

of course……

No company in the world can achieve 100% self-owned mobile phone SoC.

Especially in the CPU architecture and instruction set of mobile phone SoC, because all of them use arm.

but……

When Fang Nian established Baize, he felt that Baize should challenge the complete self-reliance of the most core components.

100% impossible, strive for 95%.

For example, the system memory particles are made in China.

Others are nothing more than some protocol standards, such as Wi-Fi with patent fees, such as mobile communication standards.

It doesn't matter, frontier academics are in the accumulation period in these aspects.

Even Wi-Fi, because with the development of the times, the standard of WiFi is also iterating...

The atmosphere in the conference room finally relaxed a lot.

Academician Chen also added: "From next week, some members of the team will enter the stage of technology sorting and integration, and will systematically sort out all the outputs in the past few months."

Fang Nian didn't say much, and glanced at the projection: "I heard that some mobile phone SoCs are beginning to integrate security chips, can they also be added?"

Academician Chen pondered: "I have heard of it, but it is not mainstream yet. The time window for the first generation is too tight, can it be delayed?"

Seeing this, Fang Nian made a gesture, chuckled and said, "Just asking to confirm."

"Although I don't know the professional knowledge inside the chip, I have some ideas after seeing this picture;

Mobile phone SoC is different from computer CPU, it is a highly integrated system chip, in order to achieve more and more functions, will it gradually iterate some new functional modules with the development of the times?

For example, audio and video with a higher experience;

There may even be new processing units, such as artificial intelligence processing units;

I think the era of mobile intelligence has just begun, and there are many unexplored areas. It does not necessarily mean that the West is the absolute standard. "

Finally, Fang Nian looked at the crowd and said with a faint smile, "Academicians, you can think about these issues carefully."

After listening to Fang Nian, all the R&D engineers present, including Academician Chen, were in a daze.

From the aura pressure brought by Fang Nian's simple two sentences, to the things mentioned in an understatement now...

They finally realized why the actual helm of a system as large as Frontier is such a young man.

Finally, of course, we also talked about the tape-out plan.

Professor Zhang eloquently said: "The original plan was to go to Gulf stream this afternoon. Considering that only TSMC currently has a 28nm process, and it is not fully mature, the result of the first tape-out may not be the same. will not be as designed;

……

The overall planning cycle from tape-out to test end is 10 days, with 1 day of redundancy before and after;

......"

Finally, Professor Zhang also talked about the crucial tape-out cost: "Because it is the latest process, only the full-mask solution can be used. The quotation given by TSMC is US$950 million once."

After Professor Zhang finished speaking, Shen Demin added: "In terms of the cost, we have negotiated with TSMC many times. It is a discounted price after considering the acceptance of the new process. There is no way to lower it."

After listening, Fang Nian smiled, and said, "Under ideal circumstances, the tape-out fee would also need 2000 million US dollars..."

"In this way, let me make it clear that at most three tape-outs can be performed under the current design of this performance standard;

After that, you have to come up with a feasible solution no matter what. The final date is June 6th. I only look at the results. Is there any problem? "

Academician Chen, Professor Zhang and others quickly answered seriously: "No problem!"

"Thank you Mr. Fang for your strong support for our experiment."

"..."

It was 3000 million yuan, more than [-] million yuan for the tape-out fee, and this kind of large-scale expenditure was also the first time in their working experience.

To know a large number of chip design teams, companies, laboratories, etc., it may take two or three years to tape out once.

The key point is not low-process tape-out, just three and a half million dollars at a time.

The $950 million tape-out cost is three times more expensive than the 45nm tape-out cost!

Even for the sake of money, they have to come up with something.

After all, with the massive accumulation of knowledge and capital investment, there are three opportunities for actual revision. If it is not enough, Academician Chen feels that he is shameless.

"..."