Chapter 211 The Only Problem Is the Cost
Although Jiang Tao had left Yuanxin two years ago, organizationally, he had never placed himself outside of Yuanxin's system. Seeing Su Yuanshan seriously mention two matters, Jiang Tao also grew serious.
"Go ahead," he said.
Su Yuanshan smiled. "In a few days, Senior Brother Gao Xiaodi will be bringing the entire Yuanxin CPU team over."
"What the heck?" Jiang Tao leaned forward instinctively upon hearing this.
Then he realized he had overreacted and chuckled, saying, "That guy coming here will definitely stir things up."
Su Yuanshan laughed heartily.
Jiang Tao and Gao Xiaodi had been classmates, with Gao Xiaodi even sleeping in the bunk above him.
Back when Su Yuanshan jokingly suggested Gao Xiaodi come over to work on cache memory, Gao's excuse had been that he didn't want to come because he felt awkward "bullying" Jiang Tao.
"Hehe, once they're here, the cache team will merge directly into the CPU team, returning under Yuanxin's architecture. After that, you'll be separate again," Su Yuanshan said.
Jiang Tao nodded naturally at first but quickly realized something.
He still had people downstairs!
Which meant dozens of people currently working in the cache team were technically employees of Yuanzhong Microelectronics!
This needed some explaining.
—When Su Yuanshan first started working on CPUs, he only had about seventy to eighty people. Apart from a dozen or so who had followed Li Mingliu to Chengguang Institute, the remaining fifty or sixty were all Gao Xiaodi's "new recruits."
With such a small team, Su Yuanshan had to juggle architecture design and cache memory development—it was almost impossible. So he had the idea to "borrow" manpower from Yuanzhong.
He asked Li Mingliu to take only a handful of key team members to Yuanzhong, then siphoned away half of Yuanzhong's staff.
At that time, Yuanzhong barely had a hundred people, and this was a significant sacrifice.
Now, with cache completed, Jiang Tao had been eagerly awaiting the return of his people—veterans polished by Li Mingliu's training, who had become priceless assets.
Jiang Tao immediately grew nervous. "When are you giving my people back?"
Li Mingliu casually tossed him a cigarette, laughing, "Old Jiang, don't be so stingy!"
"Come on... Xiaoshan, you said back then this was a joint development," Jiang Tao said, staring at Su Yuanshan. "Their salaries are being paid from my budget!"
Su Yuanshan felt a little guilty under Jiang Tao's gaze and scratched his head, chuckling. "Didn't we give you joint development subsidies? That's more than enough to cover salaries."
"..."
"Senior Brother, honestly, I can't return them," Su Yuanshan said, knowing he was being a little shameless. "How about we negotiate a compensation plan?"
"What plan?"
"We'll discuss that later," Su Yuanshan said, laughing. "First, let's talk about the second matter."
"..."
"The second thing is that Old Zhu's new team is focused on flash memory—specifically NAND flash. You know that, right?"
Jiang Tao nodded.
Su Yuanshan continued, "The application potential for flash memory is obvious. The only reason it hasn't taken off yet is the high cost."
"But we know a basic truth about semiconductors: any challenge related to cost is a challenge that can be overcome—or at least is worth overcoming."
"Therefore, I hope you can also pull together a small team to focus on developing a flash memory controller chip."
Su Yuanshan looked seriously at Jiang Tao. "No need for me to explain how flash memory control works, right?"
At this, Li Mingliu chuckled, "Better explain it just in case."
Jiang Tao shot him a glare. "What do you take me for? A flash memory device stores electrons through doping layers in a MOSFET. Even after power-off, the charge state remains. Then, you design a chip to read the 0/1 states of the storage cells, retrieve the data, and manage read/erase operations through voltage control."
"Finally, the controller interfaces with the computer's protocol..."
Su Yuanshan looked at Jiang Tao with genuine admiration. "Senior Brother, when did you learn all that?"
"...Since Old Li came back and started working on this. I figured I'd better get familiar," Jiang Tao said, lighting the cigarette Li Mingliu had tossed him.
Su Yuanshan smiled and nodded. "The principle is actually simple—no matter how you slice it, it's all about MOSFETs."
"But," Jiang Tao said, "the cost is still the killer."
"Which is exactly what we have to solve," Su Yuanshan replied with a grin.
...
At this point in time, NAND flash was still in its infancy. Commercialization was a distant dream.
The biggest hurdle? Cost.
The existing architecture was SLC—Single Level Cell. Each memory cell could store only 1 bit of data: either a 0 or a 1.
Its advantages were clear: fast, stable, and long-lasting—up to 100,000 full erase/write cycles theoretically.
But its downsides were equally obvious: low capacity and high cost. In the future, SLC would only be used in extreme industrial environments or scenarios where cost didn't matter.
Cost, historically, has been the biggest obstacle to any technology's mass adoption—and the most surmountable one.
A few years later, Intel would introduce MLC—Multi Level Cell.
Compared to SLC, each MLC cell could store 2 bits of information.
However, the manufacturing process became more complicated, and the reliability, read/write speed, and lifespan all decreased dramatically.
Still, MLC was cheaper than SLC, though it didn't immediately popularize flash memory.
True mass adoption came with TLC—Triple Level Cell—storing 3 bits per cell.
It required even more delicate voltage control and more complex processes. Its downsides remained the same—slower, less reliable, shorter lifespan—but its cost made it affordable to the masses.
Eventually, even QLC appeared—storing 4 bits per cell, even cheaper and with even shorter lifespan.
Moreover, as someone from the future, Su Yuanshan knew that while flash architecture affected read/write speed, it wasn't the only factor—or even the biggest one.
The real bottleneck was the data transfer protocol standard, followed by controller chip quality, and finally, the memory cell architecture.
For instance, the soon-to-be-born USB 1.0 protocol—aggressively pushed by Yuanxin and Xinghai—would offer a "full speed" of only 48Mbps. Although that was four times faster than the historical 12Mbps standard, it was still slow.
No matter how advanced the controller or flash memory was, under that bottleneck, performance would suffer.
Thus, the future of flash memory would require simultaneous breakthroughs in protocols, controllers, and memory architectures.
...
This trip, Su Yuanshan's main mission was to point Li Mingliu's team toward two goals.
First: move from single-layer to multi-layer structures—packing more transistors into the same area.
Second: MLC—achieving greater data density per memory cell.
Coincidentally, Su Yuanshan was deeply familiar with the foundational technologies and theories behind both directions.
The problem was, he didn't know how to subtly guide Li Mingliu's team toward these paths without outright handing them the answer.
Just like with his senior brothers, Su Yuanshan saw Li Mingliu and his peers as Yuanxin's cornerstones, destined for great things. Unless absolutely necessary, he would never simply hand them solutions and wait for applause.
That would do no good for Yuanxin's long-term future.
Thank you for the support, friends. If you want to read more chapters in advance, go to my Patreon.
Read 20 Chapters In Advance: patreon.com/Albino1