If you cheat, money will follow

R & D Center.

Mr. He and the designers connected the Changning Storage large-capacity solid-state drive brought by Mr. Zeng to the computer.

Copy the design drawing into a professional computer.

The designer opened the design drawing, looked at it for a while, and immediately exclaimed:

"So complete?"

"This is already a complete blueprint."

"Huh? It's not an ARM architecture? It's not a RISC-V architecture either? It's a brand new architecture!"

"Do you have the chip verification report?"

"Doesn't that mean we just need to tape out?"

Mr. Ren and Mr. He looked at Chang Le in surprise.

Chang Le smiled and said, "Why are you looking at me like this?"

The essence of a chip (no matter any chip) is an integrated circuit made on a semiconductor substrate that can implement specific functions.

This is a process that turns sand into gold.

In this process, design is the source and foundation.

Just like building construction, architectural design is the first step.

You first have to tell the construction staff, what is this house for?

What structure is used?

How many layers are there?

What is the internal layout?

And this set of chip design has such a function.

It has clarified the purpose, specifications, and performance of the chip - well, it is an epoch-making SOC chip.

It also has a complete circuit diagram, circuit detail design diagram, chip layout diagram, chip wiring diagram, and corresponding verification simulation results.

It also uses an unprecedented instruction set architecture, with complete chip architecture diagram and module division.

Chip architecture is similar to building architecture.

The latter usually include wood structure, steel structure, brick-concrete structure, frame structure, etc.

There are four main types of the former: X86, ARM, RISC-V and MIPS.

X86 architecture is a desktop-level chip architecture.

Designed by Intel Corporation in 1978.

On this basis, AMD developed the X2003-86 architecture in 64.

IBM's instruction set is also extended based on the X86 architecture.

The ARM architecture, developed by ARM in 1983, is a 32-bit simplified instruction set architecture.

Mainly suitable for mobile communications field, low power consumption and low cost.

Major customers include Fruit, Google, IBM, Juchang, and Samsung.

After Juchang encountered sanctions, it decisively converted to the RISC-V architecture and developed its own instruction set.

RISC-V architecture is an open architecture also developed based on reduced instruction set computing.

Completely open source, simple in design, modular in design, and complete in industrial chain.

This structure is operated by the foundation and has 275 members. Most domestic Internet companies and chip companies use this structure.

Such as Cat Factory’s Xuantie AI chip, etc.

The MIPS architecture...is dead.

"Mr. Chang, what structure did you use for this design?" asked Mr. He.

"Don't ask me, how do I know this?" Chang Le said:

"I heard that it is a new architecture developed based on the RISC-V architecture and is suitable for carbon nanotube chips. You can call it the Lingxi architecture."

"Developed based on RISC-V? It doesn't look like it?" Mr. He said:

"Lingxi structure?"

"Mr. Chang, the institute has recruited a new chip design team?" asked Mr. Ren.

"No! We do not carry out chip design business." Chang Le shook his head and said.

"Then where did this design come from? You always know. This design team is very extraordinary." Mr. He said.

"Without a team, you can just assume it is provided by artificial intelligence Lingxi." Chang Le said.

Mr. Ren obviously didn't believe it: "Lingxi?"

Mr. He: "..."

Designers: "..."

The internal version of the Lingxi large language model is also used by Juchang, after all, they are also shareholders.

but.

The staff of Juchang are mainly used for code correction, debugging, and streamlining. They can copy and paste code that is not too complicated.

As for outputting the entire chip design through this model?

Just kidding, I can’t even think about it.

It's not that powerful yet.

If you were to be so powerful, you would laugh even in your dreams.

After listening to Chang Le's words, Mr. Ren smiled and said, "Mr. Chang, do you think I believe it or not?"

Chang Le also smiled and said, "Believe it or not, this is a fact."

"Okay. Then we will treat this as provided by Lingxi. Therefore, Tingbo, this architecture will also be named Lingxi Architecture from now on, respecting other people's intellectual property rights." Mr. Ren said.

"Okay!" Mr. He nodded.

Mr. Ren originally wanted to dig out some valuable information from Chang Le and finally determine who was the team behind it?

where?

Then the check was waved and the entire establishment dug in forcefully.

In addition to Leda Investment, there is nothing that other teams cannot find.

"Mr. He, what are you going to do next?" Chang Le asked.

"I suggest you verify it again just to be sure. If you tape it out directly, if it fails, the cost will be too high," Mr. He said.

Taping out, in short, means debugging equipment and trial production.

A 14nm process chip costs about US$300 million per tapeout;

A 7nm process chip costs about US$3000 million per tapeout;

For a 5nm process chip, one tapeout can cost up to 5000 million US dollars.

"Verify again? Where to go? Do you verify it yourself?" Chang Le asked.

"No, go to Huahong Semiconductor. They also have their own integrated circuit testing business. Once successful, you can directly enter the tape-out stage with them to save time." Mr. He was a little excited.

"Mr. Chang, according to this design, if the tape-out is successful, it may be an epoch-making chip." Mr. He said to Chang Le:

"As far as chips are concerned, regardless of other factors, all SOC chips and desktop chips from other companies will enter history."

"No wonder, a few days ago you called Mr. Ren and told him that our design idea was wrong..."

"After looking at this set of designs, we have indeed fallen into the misunderstanding of silicon-based chips."

"Of course, the success of the design is only a theoretical success."

"There are other factors to consider in specific production. If the process is not up to standard, it will also lead to a decline in chip performance."

This point can be referred to Samsung's 5-nanometer chip, which is very impressive.

"What's the progress of Changxin's DRAM?" Mr. He asked the R&D staff.

"It's not ideal and the progress is not fast." said the R&D staff.

"Oh, Mr. Ren, I suggest that we send a team to assist them in research and development, and at the same time, we can also do a good job in matching." Mr. He said.

"Yes. The chip is strong, but if the memory is not strong, it will be a serious drag." Mr. Ren said:

"Mr. Chang, Mr. Zeng, come to my office to talk? We don't understand professional issues. Let professionals solve them."

"Let's talk about something else."

"Okay, but first let's talk about it. Our institute really doesn't have a design team to recruit." Chang Le said.

"That's not it. Let's talk about something else." Mr. Ren said.