1991 Starting from the core

Although he has left Yuanxin for two years, Jiang Tao has never placed himself outside Yuanxin in terms of system. Seeing Su Yuanshan talking about two things in a very formal manner, he also became serious.

"Well, you say."

Su Yuanshan smiled: "In a few days, senior brother Gao Xiaodi will bring the entire CPU team of Yuanxin over."

"Shall I wipe it?" When Jiang Tao heard this, his neck immediately leaned forward.

Then he immediately realized that his reaction was too big, and smiled and said: "That guy will make a lot of noise when he comes over."

Su Yuanshan laughed.

Jiang Tao and Gao Xiaodi were classmates, and Gao Xiaodi was even the brother who slept in his upper bunk.

Su Yuanshan joked with Gao Xiaodi and asked him to come over to do caching, but Gao Xiaodi's excuse was that he couldn't come and was too embarrassed to bully Jiang Tao.

"Haha, after being brought over, the cache team will be directly merged into the CPU team, and then return to the far-core architecture, and it will be separated from you in the future."

Jiang Tao nodded naturally, but immediately realized a problem.

He still has someone downstairs!

In other words, there are still dozens of people in the cache team who are from Yuanzhong Microelectronics!

Then, it’s a long story.

——When Su Yuanshan started working on CPU, there were only 70 or 80 people in total. Among the seventy or eighty people, except for the dozen or so people who had been following Li Mingliu to support Chengguang Institute, the remaining fifty or sixty people were all "new soldiers" led by Gao Xiaodi.

With just a few people, Su Yuanshan has to build the architecture and cache, which is really a bit difficult to uncover. So Su Yuanshan set his sights on Yuanzhong Microelectronics - he asked Li Mingliu to come to Yuanzhong with only a few core team members, and then directly pulled half of Yuanzhong's people away.

You know, Yuanzhong was only a hundred or so people at that time, and they were gritting their teeth to support Su Yuanshan.

Now that the cache is done, Jiang Tao is still eagerly waiting to return the favor - those veterans who have been tempered by Li Mingliu are real treasures.

So Jiang Tao immediately became a little nervous: "When will I return my people?"

Li Mingliu threw a cigarette to Jiang Tao and said with a smile: "Old Jiang, don't be so stingy!"

"Oh my god... Xiaoshan, what we were talking about at the beginning was joint development." Jiang Tao stared at Su Yuanshan and said, "I pay all their wages!"

Su Yuanshan was a little embarrassed by Jiang Tao, and he chuckled: "Didn't I give you a joint development subsidy fund? It's enough to cover your salary."

"..."

"Brother, I really can't pay you back." Su Yuanshan knew that what he did was a bit unkind, so he thought for a moment and said, "Otherwise, let's discuss a compensation plan."

"What plan?"

"Let's discuss it later." Su Yuanshan chuckled and said, "Let's talk about the second thing first."

"..."

"The second thing is that the new team established by Lao Zhu specializes in flash memory. To be precise, it specializes in NAND flash memory. You should know about it."

Jiang Tao nodded.

Su Yuanshan said: "That's right. There must be applications for flash memory, but they haven't been launched yet. The only reason is that they are too expensive."

"But we have basic common sense that... in the field of semiconductors, any problem caused by high prices can be broken through, or is worth breaking through."

"So I hope that you can also deploy a small team to cooperate with the main control chip."

Su Yuanshan looked at Jiang Tao as he spoke: "You don't need me to explain the control mode of the flash memory, do you?"

Here Li Mingliu laughed again: "You'd better say it."

Jiang Tao glared at Li Mingliu: "Who are you looking down on? A MOSFET (MOS field effect transistor) preserves electrons through the doping layer and can maintain its state even after power is turned off. This is the basic principle of memory. Then design a chip, By obtaining the 0 and 1 states of the memory, the data distribution can be obtained, and then applying voltage according to the read and erase commands can complete the data operation."

"Finally, communicate with the computer's protocol interface..."

Su Yuanshan looked at Jiang Tao in surprise: "Senior brother, you are so awesome...when did you know about it?"

"...Lao Li started doing this after he came back. Is it okay for me to touch it?" Jiang Tao rubbed the cigarette that Li Mingliu had just thrown to him on the table a few times and lit it with a click.

Su Yuanshan smiled and nodded: "The principle of this thing is actually quite simple. It can't be separated from the MOS field tube after repeated over and over again."

Jiang Tao said: "But whether the cost can be reduced is always a big problem."

"So we just have to address the cost issue."

Su Yuanshan said with a smile.

…

NAND flash memory is still in its infancy, and it is far from being truly commercialized.

The most fundamental reason is that it is expensive.

The current architecture is still SLC architecture, Single Level Cell. There are only two variations of 0 and 1 in one unit, and it can only store 1 bit of data.

Its advantages are fast, good stability, and long life-theoretically it can reach a life of 100,000 times of complete erasing and writing.

But it is also obvious that the capacity is low and the cost is high. In later generations, the only scenarios where it can be used are industrial fields in extreme environments and other fields regardless of cost.

However, cost has always been the biggest obstacle to the popularization of a technological product, and it is also the easiest obstacle to overcome.

So a few years later, Intel developed a new architecture, MLC—Multi Level Cell.

Compared with SLC, each cell unit can store 2 bits of information.

Correspondingly, its manufacturing process is more complicated, while its reliability and reading and writing speed are also reduced, and its service life is exponentially reduced.

But even so, its price is still slightly expensive and has not brought real popularity to flash memory.

What really made flash memory popular was actually TLC—Trinary Level Cell, which translates to triple unit.

Each cell unit can store 3 bits of information and requires more complex processes and voltage control.

The shortcomings are the same as always.

However, it is the absolute contributor to bringing flash memory into thousands of households.

As a result, QLC was born later - each cell unit can store 4 bits of information, with a larger capacity, but slower, shorter lifespan, but the cost can also be reduced to a lower level.

Moreover, Su Yuanshan, who came from a later generation, knew that in the development process of flash memory products, the architecture of flash memory is only one of the factors that restricts the read and write speed of the product, not even the biggest factor.

The biggest factor is actually the interface protocol standard for data transmission, followed by the main control chip and related technologies, and finally the architecture.

Just like the upcoming USB 1.0 protocol, although a relatively radical solution was adopted under the initiative of Yuanxin and Xinghai, the transmission rate at full speed is only 48Mbps - which is already four times the historical 12Mbps.

Under this protocol, even if you use the most powerful main control chip and the most powerful flash memory particles, it will still be slow.

Therefore, the future of flash memory must rely on protocols, master control, and granular architecture to advance simultaneously.

…

Su Yuanshan came here mainly to point out two directions to Lao Li.

The first direction is to turn a single-layer structure into a multi-layer structure, so that as many transistors as possible can be stacked per unit area.

The second direction is naturally MLC.

Coincidentally, Su Yuanshan knew all the basic technologies and principles related to these two directions.

But what was irritating was that he didn't know how to tactfully guide Li Mingliu and the others on the right path.

Just like his attitude towards his senior brothers, Li Mingliu and the others are also the cornerstone of Yuanxin and have great uses. Unless it was absolutely necessary, Su Yuanshan would not be able to personally propose a plan and then wait for these people to applaud him.

That won't do the telecore any good.