Chapter 336 The Pentium Flaw
The two Saul-architecture CPUs undergoing evaluation both had a core frequency of 75MHz and utilized the most advanced 0.5-micron manufacturing process of the time. They integrated Yuanxin's newly developed enhanced multimedia instruction set, MMX, and thanks to the 0.5-micron process, were able to feature 64KB of full-speed L2 cache—operating at the same frequency as the core itself.
Previously, there had been intense debate between the teams over whether to integrate L2 cache, and if so, how large it should be and whether it should run at full or half speed. Ultimately, it was Su Yuanshan who made the final call, deciding that Starsea's CPUs should take a more moderate, market-oriented approach.
In Su Yuanshan's long-term plan, Starsea's CPU division was intended to be the main force, alongside AMD, to challenge Intel's throne over the coming years. Therefore, they needed to prioritize market acceptance. Yuanxin's internal CPU team, on the other hand, would be developed based on circumstances—ideally securing full licensing for Intel's x86 instruction set by the following year, allowing for a mutually beneficial situation.
If that failed, they would pursue a "curveball rescue" approach: after Starsea's IPO, they could use the raised capital to establish a new joint venture in mainland China with Yuanxin and Texas Instruments, dominated by Starsea, to roll out CPUs designed by Yuanxin.
The worst-case scenario?
Yuanxin's CPU division would pivot into a pure research unit, hibernate for five years, and then leap directly into developing 64-bit processors.
Given this strategic setup, it wasn't surprising that Su Yuanshan valued Cyrix more heavily than even the CPU team at the Shanghai tech park.
...
The cameras faithfully recorded the progress of the testing.
As the real-time test results rolled in, the designers and testers alike displayed growing looks of disbelief.
"The impact of full-speed L2 cache is... this huge?" asked Holpes, Claude's deputy and the longtime liaison with Li Mingliu's team, unable to contain his excitement halfway through the tests.
"It's probably due to improved instruction hit rates, dramatically boosting execution efficiency," Claude replied, keeping his face composed but equally unable to hide his joy. "Plus, the overall architecture itself is noticeably more advanced compared to Pentium."
Although simulation and design calculations had already given a rough idea of the CPU's capabilities, there was an old saying in chip design: "Everything is theory until you tape out."
Until the chip was physically manufactured and tested, all assumptions were just guesses.
"Right now, it looks like the biggest issue with this CPU will be cost," Claude said as he pulled out his phone to call Su Yuanshan.
Testing was more than halfway through. On the Quake benchmark, the new CPU paired with a custom motherboard had achieved near-legendary results—almost doubling the FPS compared to Intel's best Pentium 100MHz chip.
That's right: the Saul CPU, to ensure Socket 7 user compatibility, hadn't switched to a new socket but stuck with Socket 7, allowing it to run on existing Pentium-supporting motherboards.
However, for optimal performance, it still needed the chipset co-designed by SOYO and Yuanxin.
"Claude, something's wrong,"
An engineer pointed at one of the computers, frowning. A batch of floating-point division tests had just finished on several systems.
This test, provided by Yuanxin, simulated real-world engineering, mathematics, and other professional scenarios through heavy floating-point operations.
The correct result across platforms was a precise value for Pi to five decimal places: 3.14159.
But one of the systems—the one running on an original Pentium 66MHz—had spat out a wildly wrong answer: -2021.211.
"Run it again?" Claude put away his phone, intrigued, and leaned in.
After repeating the test, the same wrong result appeared.
Everyone exchanged glances. Claude even forgot about calling Su Yuanshan for a moment.
After a few seconds of thought, he spoke carefully, "If the same software produces correct results on five different platforms, and only this one shows errors... there can only be one explanation."
"This test is a floating-point division benchmark handled directly by the CPU's FPU.
Clearly, this Pentium 66's floating-point unit is defective."
Claude quickly regained his engineering sharpness.
He instructed, "Grab more early-generation Pentiums. Let's see if this is an isolated issue."
...
The engineers rushed off and soon returned with several Pentiums launched last year, ranging from 33MHz to 66MHz models, and even some later second-generation Pentiums.
For extra caution, they reran the tests across all machines.
The conclusion was clear: Yuanxin's floating-point division testing tool produced a consistent, single correct answer.
And that 66MHz Pentium machine? It stubbornly returned the same -2021.211 error every time.
Tension filled the room.
Everyone present was an experienced CPU engineer.
They all knew that if a CPU failed calculations, it usually resulted in system crashes or random output—not consistent, repeatable errors.
A consistent, reproducible flaw implied a fundamental design defect.
Seconds ticked by.
When they finished testing the new batch of CPUs, everyone's faces froze.
Every first-generation Pentium produced the wrong result!
"Fuck... Did Intel really make such a mistake?" Holpes stared wide-eyed, trying to hide the excitement brimming inside him.
As a veteran who had been chasing Intel since Cyrix's early days, he knew how formidable Intel was in both design and manufacturing.
"Maybe," Claude said, quickly calming down.
A flood of possibilities flashed through his mind, but finally he fixated on one idea—perhaps the error lay in the division lookup tables.
At that time, to speed up floating-point division, every chip company—including Cyrix—had hard-coded division tables directly into the CPU core.
This method had obvious advantages for speed, but it also carried a hidden risk: these tables were manually crafted, not automatically generated by EDA tools.
Thus, they couldn't be fully simulated during design.
Mistakes were entirely possible.
Could it be that Intel had screwed up their division tables during design?
Realizing the stakes, Claude immediately grabbed the phone again and dialed Su Yuanshan's number.
Right now, reporting the good news about Saul's success was secondary.
Uncovering the exact flaw triggered by Yuanxin's floating-point testing software had become the top priority!
Get 30% off on my Patreon and enjoy early access to new chapters.
You can also purchase the next 100 chapters of the novel directly from my Patreon page.
Hurry up! The promotion ends on March 2, 2026.
Read 30 Chapters In Advance: patreon.com/Albino1
Chapter 336 A Good Day
At the Yuanxin Technology Park, Su Yuanshan was reviewing the experimental design report submitted by He Chunhua.
Since he had already decided to groom this "external senior brother" into the future lead of the graphene laboratory, he had to hold him to the highest standards—even if it seemed a bit ridiculous.
This was He Chunhua's third experimental design in just two weeks. If it weren't for the fact that Su Yuanshan had been carefully reviewing each version and providing thoughtful, detailed feedback, He Chunhua would have surely thought Su was deliberately making his life difficult.
It wasn't until He Chunhua arrived at Yuanxin that he truly realized just how extraordinary that "junior student" who once came to learn in Peking University's lab really was.
No exaggeration—Su Yuanshan was the spiritual pillar of Yuanxin.
And curiously, within Yuanxin, no one called him a "genius." It was as if that word wasn't enough to describe him. Instead, in private, people referred to him as "Mountain God."
A man who existed like a deity—although admittedly, the nickname sounded a little silly.
"Mm, looks good. Thank you, Senior Brother."
After half an hour, Su Yuanshan finally looked up from the report and smiled. "Roughly speaking, if you follow through with this design, you might discover performance traits beyond just the Hall effect."
He Chunhua exhaled deeply.
After all, he was a PhD in physics from Peking University. No matter how brilliant Su Yuanshan was—even if he was borderline godlike—he still had his own dignity.
His experimental design wasn't just about replicating Su Yuanshan's Hall effect results. It was more about building upon that experiment to explore further directions.
Since Yuanxin could provide the top domestic research environment and the most generous international-grade funding, it would be a waste to merely follow Su Yuanshan's footsteps. That would be a betrayal of everything his mentor, Professor Li Chungang, had taught him.
He stood up, fired up. "Then I'll start assembling the team this afternoon!"
"No need to rush," Su Yuanshan said, glancing at his watch. It was already past eleven. He stood up and said, "Come on, let's visit the lab together, chat a bit, and grab lunch."
Just as he spoke, his private phone rang.
This wasn't a line just anyone could call—it was reserved for family, friends, and a few key department heads for emergencies.
Seeing the caller ID, Su Yuanshan froze for a second, then a smile spread across his face.
"Hold on, Senior Brother. I think this is good news."
He answered the call. As expected, it was Claude's voice.
"Su, tape-out was successful. We just finished testing against our competitors. We're ahead across the board."
"Excellent work!" Su Yuanshan praised, exhaling a long breath of relief.
They had used a Pentium MMX-based architecture, even incorporating ideas from the legendary Tulatin core, plus full-speed L2 cache. If this thing couldn't beat the current Pentium 1, he might as well quit the game.
But no matter how great a chip design looked on paper, the real test came after fabrication. Hearing that the results met expectations across the board lifted a huge weight off his shoulders.
"Still," Claude said cautiously, "based on fabrication data from TI, the unit cost is a bit high. And considering Cyrix's past pricing and brand image..."
He paused, but Su Yuanshan immediately picked up on the subtext—Claude was worried about whether the market would accept it.
Su Yuanshan chuckled. "Don't worry. If Intel released this chip, it might struggle to sell. But with Starsea's name on it, it will absolutely sell. And trust me—Intel will follow our lead and start using full-speed L2 cache."
"Why?" Claude asked, curious.
Su Yuanshan knew Claude was a pure tech guy—someone who believed in beating the market with superior products, just like Intel once did. The idea that marketing and positioning played a bigger role than specs was probably foreign to him.
"Simple," Su Yuanshan explained. "The market needs a serious competitor to Intel. So does capital. As for Cyrix's old pricing reputation, sure, it has some effect—but not much. The new CPU will be launched under Starsea's name and a new trademark. From now on, the Cyrix brand will handle the mid-to-low-end product line. As for Intel—do I even need to explain? Full-speed L2 cache is the future. If they don't adopt it for their high-end CPUs, they'll be digging their own grave."
"Got it," Claude replied quickly. Ever since Cyrix had been absorbed into Starsea, its old sales and marketing strategies had been scrapped. Starsea had taken full control. His question had been more of a formality.
Su Yuanshan, however, couldn't help but wonder—did Claude seriously not notice the floating-point division flaw in the first-gen Pentium? That would be surprising. He had emphasized from the beginning that every major CPU on the market—especially Intel's Pentium line—should be benchmarked for performance ranking and pricing strategy.
Had they skipped over testing the first-gen Pentium? Or forgotten to use the scientific calculation test tool Yuanxin had provided?
Claude's next sentence cleared up Su Yuanshan's doubts.
"By the way… during the floating-point division tests, we found something odd," Claude said, hesitating slightly before continuing.
Su Yuanshan grinned.
"What kind of odd?" he asked.
"It seems… Intel's floating-point division might have a flaw," Claude said, before sharing his full hypothesis.
Su Yuanshan listened quietly. When Claude mentioned wanting the source code for the testing tool, he nodded. "The source is already on StarHub. Just download it yourself."
Claude took a deep breath. "If… if we confirm that Intel's first-generation Pentium has a design flaw, what should we do?"
Su Yuanshan narrowed his eyes and smiled. "Tell Carly. She'll know what to do."
"Understood!"
...
After hanging up, Su Yuanshan let out a long exhale.
If his own internal hang-ups prevented him from launching a direct attack on Intel, then Carly could make the move.
And in the midst of preparing for an IPO, there was no way she'd pass up this golden opportunity to strike Intel while it was weak.
"Let's go, Senior Brother. Lunch is on me!" Su Yuanshan called out with a laugh to He Chunhua, who had been listening intently.
"From what I just overheard, sounds like Starsea found a flaw in Intel's chips?"
"Shhh… that's confidential."
As Su Yuanshan walked out of the office, he opened his arms to embrace the crisp autumn breeze and exclaimed softly—
"What a good day."
Get 30% off on my Patreon and enjoy early access to new chapters.
You can also purchase the next 100 chapters of the novel directly from my Patreon page.
Hurry up! The promotion ends on March 2, 2026.
Read 30 Chapters In Advance: patreon.com/Albino1
Chapter 338 The Most Direct Competitor
He Chunhua returned to the laboratory office with his project proposal in hand. The Yuanshan Laboratory had now been officially established, and two groups from T University and the Provincial University had already started their experiments, each bringing their own topics and participating through the educational research pathway.
Meanwhile, He Chunhua's team had joined directly under Su Yuanshan's project group, meaning they worked more closely with and under the supervision of Su Yuanshan himself.
"Old He, how did it go?"
Seeing He Chunhua return with a cheerful look, Xia Xiaoxuan immediately stood up, unable to contain her excitement.
Xia Xiaoxuan was a combined master's and PhD student in material physics at Peking University—and notably, she was a woman.
Her supervisor was none other than the department head, Professor Xu Zhiwen. Su Yuanshan had met her once during a visit to Xu Zhiwen at Peking University.
Originally, Professor Xu intended to join Su Yuanshan's open lab by bringing his own research project, but unexpectedly, Professor Li Chungang beat him to it by assigning his students directly to Su Yuanshan's group.
With that move, Xu Zhiwen dropped the educational research pathway idea—after all, Peking University was Su Yuanshan's alma mater. If other schools wanted to join, they wouldn't even have the opportunity.
So Xu Zhiwen gritted his teeth and sent Xia Xiaoxuan, along with a few junior students, to directly enlist under Su Yuanshan's team.
Though from different fields, within Su Yuanshan's lab they were all "Peking University people," so there was no distinction.
Since Su Yuanshan had made it clear that it wasn't the right time to move too aggressively, the group had simply formed one large project team to work on extending Su Yuanshan's previous graphene research.
This wasn't just strategic—it was targeted.
Having Su Yuanshan, the discoverer of graphene, listed as a co-author or corresponding author on their papers significantly boosted their chances of publication.
"Starting work tomorrow!" He Chunhua announced proudly.
"Nicely done!" Xia Xiaoxuan beamed.
Though she looked like a gentle, ponytail-sporting university girl straight out of an MV, she was highly driven.
She had grown impatient over the past few days, having had to wait and assist He Chunhua in completing the formal paperwork.
Still, she couldn't help but grumble, "Mountain Boss is taking his sweet time. Isn't he worried someone else might steal the thunder..."
"Hehe, other than us, who's even aware that graphene can exhibit the Hall effect?" He Chunhua replied, fully confident. "Even if someone else suspected it, they wouldn't know how to fabricate the Hall devices."
"I've seen him conduct experiments myself—his rigor far exceeds that of most graduate students. You'd never guess he was only a sophomore then."
"Alright, alright, we get it—you two shared a lab," Xiaoxuan teased.
"..." He Chunhua could only roll his eyes.
"By the way, Mountain Boss suggested we both design separate experiments to submit simultaneously. Maybe we can even publish back-to-back papers!"
Xia Xiaoxuan froze for a second, then cheered. "Excellent! I love that idea!"
Just as she said this, a refined-looking young man walked into the office accompanied by two graduate-student-looking assistants.
He glanced around and asked, "Is the person in charge here?"
He Chunhua, who had his back to the door, turned around. Seeing who it was, he instantly stood up straight and greeted respectfully, "Professor Xi, how can we help you?"
"I was just passing by and thought I'd stop in for a look," Xi Xiaoding said warmly.
He gave He Chunhua a gentle smile and nodded politely toward Xia Xiaoxuan. "When does your team start?"
"Tomorrow morning. We're wrapping up some preparations today," He Chunhua replied.
"Good." Xi Xiaoding nodded and added as he was about to leave, "One reminder: when using the etching machines for nanoscale operations, try to schedule for nighttime. Otherwise, you might end up like someone here... doubting your whole life after repeated failures."
"Uh... thanks for the advice, Professor Xi," He Chunhua said quickly.
Xi Xiaoding smiled again and left the office.
Of course, the "someone" he was referring to was Su Yuanshan—
Back when Su Yuanshan had secretly tried to fabricate Hall devices, repeated failures led him to mistakenly blame his bad luck.
It was only later he realized the construction site across the street had been causing massive electromagnetic interference during the day.
Watching Xi Xiaoding leave, He Chunhua exhaled deeply.
"I always heard Professor Xi and Mountain Boss were close, but now I've seen it for myself—he even knows about Mountain Boss's most embarrassing failures."
Xia Xiaoxuan recovered from her awe as well.
"That was really Professor Xi? Xi Xiaoding?"
"Yup—how could you not recognize him?"
"How would I? Yuanxin's filled with big shots," Xiaoxuan shrugged.
"I recognized one of the guys with him, though. He's from our university's math department. Total genius. But standing next to Professor Xi, he looked like a sidekick..."
"Professor Xi's one of a kind at Yuanxin. No need to feel bad."
"Yeah, you're right," Xia Xiaoxuan said wistfully.
"Also... Professor Xi is really handsome, isn't he?"
"..."
...
Now that the lab had a capable team assembled, Su Yuanshan was free to shift his attention back to the wafer fab.
According to the current timeline, the fab would begin full production right after the Lunar New Year.
Once operational, with its 8-inch wafers and 0.8-micron process technology, the fab would become a literal goldmine.
Even if they only produced Yuanxin's own power management chips, MCUs, and mobile chips, it would be enough to keep the fab profitable.
But this would inevitably make Yuanxin's relationships with certain previous partners—like UMC, TSMC, and Huajing—more complicated.
Of course, Huajing was a different case.
Their position and business model didn't really conflict with Yuanxin's, and Yuanxin had genuinely helped them back when few others would.
As far as Huajing was concerned, there was no betrayal.
Plus, with China's electronics industry booming, there was room in the market for two domestic fabs.
UMC and TSMC, however... were another story.
Especially UMC.
Once Deyuan Semiconductor went into full production, Yuanxin would start gradually pulling back its outsourcing orders.
Eventually, they would compete head-to-head.
Because Deyuan wasn't just a chip designer.
It was a pure Foundry—a professional contract manufacturing fab.
And under the current industry structure, Deyuan's real competitors would be none other than UMC and TSMC.
Get 30% off on my Patreon and enjoy early access to new chapters.
You can also purchase the next 100 chapters of the novel directly from my Patreon page.
Hurry up! The promotion ends on March 2, 2026.
Read 30 Chapters In Advance: patreon.com/Albino1
Chapter 339 Old Gao's Growth
In late November, Chen Jing returned from the capital. Due to Starsea's involvement, the original two-country, four-party cooperation on the LCD panel project had evolved into a two-country, five-party arrangement, and finally into a two-country, six-party game. Regardless of whether one was an insider or an outsider, it was now clear how seriously all parties were treating the future of LCD panel technology.
Especially after the capital's government, representing national interests, got involved. Following the joint visit by the Planning Commission and the Ministry of Foreign Trade and Economic Cooperation to Yuanxin, a special National Asset Investment Management Committee was established. Using this committee as a front, the government formally entered the negotiations and partnerships.
In the end, after intense negotiation, the six parties reached several agreements: the new LCD panel enterprise would be jointly founded by Yuanxin, Sony, Fujitsu, BOE, and the newly formed National Asset Management Committee. Separately, BOE would accept an equity investment from Starsea Venture Capital. It was an outcome that left everyone reasonably satisfied.
With that settled, Chen Jing returned, and Su Yuanshan once again boarded a flight to Shanghai. He needed to personally oversee the wafer fab and also check on the CPU team—after all, although Cyrix's chip had successfully taped out and a press conference was being prepared, the local team had yet to send their chip to the fab for mass production. Even if they were waiting for Deyuan Semiconductor, the speed was a bit too slow.
...
"I was wondering why it felt so cold today—turns out it's already the start of winter," Su Yuanshan said as he walked alongside Li Mingliu down the wide road of Shanghai's tech park, pulling his jacket tighter against the chilly wind. "Why does Shanghai feel colder than the provincial capital?"
"You geography dummy, they're on the same latitude. Of course it can get colder if the wind picks up," Li Mingliu laughed. "Anyway, once we get the landscaping done, it won't feel as harsh."
"Yeah, hurry up with that. Our tech park can't end up looking like those dusty factories in the Special Economic Zones. Even if we're building a fab, it's a high-end fab. Totally different category."
"Haha, am I hearing you discriminate against regular factory workers?"
Su Yuanshan just grinned. "Think whatever you want. I didn't secure all this land for it to go to waste."
After a few more jokes, they continued on.
"I figured you'd show up around now," Li Mingliu said after a few steps. "I've been in touch with Old Gao's team. I even helped with optimizing the L2 cache protocols, so I know their real progress."
"Mm. I'm not in a hurry... but I still want to see it with my own eyes," Su Yuanshan said. "After all, on paper it's my team. I hope Old Gao can really grow into the role."
Li Mingliu dropped his gaze, pondered for a moment, then smiled. "Old Gao will rise."
"Hmm?" Su Yuanshan looked at him in surprise.
At Yuanxin, it wasn't Su Yuanshan or Xi Xiaoding that the technical core staff feared most—it was Li Mingliu. He was a recognized genius but also someone with a sharp superiority complex when it came to technical skill and intelligence.
In daily life, Li Mingliu came across more as an intensely proud tech lead than an aloof big shot.
There weren't many people within Yuanxin he genuinely admired.
At the beginning, he hadn't thought much of Gao Xiaodi at all.
But Li Mingliu had one strength—he could recognize reality.
When he realized that Gao Xiaodi was better suited to lead the CPU department, he had decisively handed over his team and moved into new fields.
Now, to hear him affirm Gao's potential...
"Old Gao's been doing well lately," Li Mingliu said again, nodding thoughtfully. "At first, I had a bias against him."
Su Yuanshan instinctively glanced up at the sky. "It's a gloomy day today, no sun out... must be an omen."
"Haha, President Shan..." Li Mingliu clearly enjoyed seeing Su Yuanshan's rare display of humor, the kind he only showed to friends. He chuckled, "Different positions, different perspectives. I realized I can't judge everyone by my own standards."
Su Yuanshan immediately caught his implication and made a face. "Tch. So you do still think Old Gao falls short of your standards."
"No," Li Mingliu corrected. "Gao's actually very skilled in CPU architecture and team coordination. Since he took over, the team's cohesion is far better than when I led it. That's his strength.
My frustration before came from the fact that he's smart but wasn't pushing himself hard enough."
"But lately... I think he's getting it." Li Mingliu narrowed his eyes slightly, smiling. "Unlike you, not everyone can glance at a new field and immediately master it."
Su Yuanshan froze for a second, then cursed with a grin. "Damn you, that's low."
"President Shan, no offense. You're a genius, whether you admit it or not," Li Mingliu said softly, looking off into the distance. "As for the rest of us... compared to average people, we're just a bit better educated and maybe a bit smarter. If we don't push ourselves, how can we ever compete with the true titans out there?"
"I think Old Gao finally realized that too. That from now on, the only opponent for his team is Intel."
"..."
Su Yuanshan sighed softly. "Pushing hard is fine, but Gao's field—CPU architecture—relies a lot on inspiration... It's not something you can brute-force."
"He's probably thinking about dual-core architectures," Li Mingliu said, shrugging. "I overheard him excitedly talking about it the other day. Sounded pretty clear-headed, not like someone chasing illusions."
Su Yuanshan froze again.
It took him a while to find words. "Damn!"
...
Walking into the newly completed tech park office building, Su Yuanshan practically rushed toward Gao Xiaodi's office.
The CPU project on Yuanxin's side had finished its major early work; now they were deep into final optimizations and refinements.
The main structure was done—the rest would be a long simulation and tuning phase.
Hence why Su Yuanshan had earlier joked, "What's there to push for now in architecture?"
The Saul architecture would serve them well for at least another three to five years, especially in an era where MHz still dominated the conversation.
At this stage, secondary cache improvements and frequency boosts would yield bigger gains than architectural changes.
And yet, here was Gao Xiaodi... already secretly working on dual-core designs—
without even telling Su Yuanshan.
Now Su Yuanshan was dying to see what kind of wild idea Gao had come up with, especially if even the famously critical Li Mingliu thought it sounded promising.
Get 30% off on my Patreon and enjoy early access to new chapters.
You can also purchase the next 100 chapters of the novel directly from my Patreon page.
Hurry up! The promotion ends on March 2, 2026.
Read 30 Chapters In Advance: patreon.com/Albino1
Chapter 340 The Vision of Dual-Core
The seventh floor of the tech park office building. In a city like Shanghai, where every square meter of land was precious, Yuanxin's tech park and wafer fab still enjoyed over two thousand acres of space, a testament to the city's concrete commitment to supporting the semiconductor industry.
Gao Xiaodi had long harbored a deep fixation on having a proper office. As one of the first batch of graduate students at Yuanxin, back when even being a college graduate made one a "proud son of heaven," a graduate degree placed one in the elite of elites. Yet Gao still had to work alongside undergraduate and vocational students in the open office area. Even after Li Mingliu departed, Gao stayed rooted among the cubicles, training the "green recruits" from scratch on basic EDA operations.
Even after taking on CPU architecture design, when he technically had the right to a private office, he still preferred to stay embedded with his team—until now.
Once the CPU project was sent off for its final fabrication run and Gao officially took charge of the CPU department, he finally allowed himself a proper office—one spacious enough to screen movies if he wanted.
Ironically, this oversized office mostly served as the architecture design team's makeshift conference room.
Inside, a bunch of engineers were smoking up a storm, quietly discussing the sketches and notes scrawled across the blackboard.
Gao Xiaodi, taking a heavy drag from his cigarette, waved a hand and said, "That's it for today. Looks like everyone's brain is fried. How about I give you guys a few days off? Go take a vacation or something?"
"Old Gao, you said it! Don't back out later," said Ma Jingcheng, lighting up immediately. "Though I'm guessing Old Li won't approve it."
Ma Jingcheng was a "sea turtle"—a master's graduate from the National University of Singapore. After completing his studies, he joined Texas Instruments' semiconductor plant in Singapore and, unsurprisingly, found "the organization"—he reconnected with the team recruited by Zhang Rujin back in China.
When Zhang Rujin returned to establish a fab in China, Ma Jingcheng followed. But rather than joining the fab directly, he opted to join Yuanxin's CPU design department, preferring the creativity of design work over manufacturing.
It turned out to be a good decision. Ma Jingcheng quickly proved himself in CPU logic design and rose to become one of the department's key figures.
"I'll give you six and a half days," Gao Xiaodi said, flicking away his cigarette. "Flexible thinking, brothers."
The office erupted in laughter.
After all, company policy dictated that vacations longer than seven days required direct approval from the tech park leadership. Anything less just needed departmental sign-off.
"You're not going too?" someone asked.
Gao shook his head. "I'm staying. Damn it, Claude's been reporting good news nonstop. We can't let them steal all the spotlight. We've got to develop some black technology of our own; otherwise, we'll be dwarfed every time we meet."
At that moment, a playful voice called from the door, "Who's afraid of being dwarfed?"
...
Three minutes later, Su Yuanshan entered the office, waiting until the cigarette smoke cleared a little before stepping inside.
"Senior Brother, are you trying to recreate the Peach Banquet?" Su Yuanshan joked, standing by the window, grimacing. "I swear the smoke density here could be measured in octaves."
"Hehe, I'll install an exhaust fan tomorrow," Gao Xiaodi said sheepishly.
Seeing Su Yuanshan standing, Gao and Li Mingliu didn't dare sit either. They casually leaned against the desk.
"Ever thought about just smoking less?" Su Yuanshan said helplessly, casting a sharp glare at them. "You lot better start researching nanorobots soon—might need them to cure your future lung cancer."
"Hehe..."
Everyone chuckled awkwardly.
Once the room cleared out enough to breathe, Su Yuanshan sat down.
He glanced at the blackboard, which displayed two square blocks representing cores, surrounded by various protocols and parameters—mostly brainstorming ideas.
"Working on dual-core?" Su Yuanshan asked, scanning the room with a smile.
"Yup," Gao Xiaodi said, scratching his head. "We ran some serious simulations. Even though we're barely hitting 100MHz now, it won't be long before CPUs break through 1GHz. Frequency scaling has limits. When that happens, dual-core—or even multi-core—will become inevitable."
"And if it's inevitable, we might as well start preparing now, so we're not scrambling later."
Su Yuanshan nodded appreciatively. "Makes sense. What's the current bottleneck?"
Gao grimaced. "Plenty. First, we're not sure whether to go with dual-die packaging or true dual-core on a single die. Then there's the internal communication architecture, data handling protocols, arbitration units... not to mention fabrication challenges. Even if we design it, today's fabs might not be able to manufacture it."
Su Yuanshan looked around. Everyone had heavy expressions—clearly wrestling with the technical and manufacturing hurdles.
After a few seconds of thought, Su Yuanshan decided he needed to guide them properly.
He was genuinely impressed and happy that Gao Xiaodi had anticipated the multi-core future so early.
But as someone who had lived through the messy birth of "glued dual-core" CPUs, he knew the dangers firsthand.
If Gao chose the dual-die (glued dual-core) route, they could end up wasting years of effort.
And Su Yuanshan didn't have the luxury of American companies to afford that kind of waste.
"We should discuss this carefully," Su Yuanshan said, walking up to the blackboard and picking up a piece of chalk.
"On paper, dual-die packaging seems easy, but it's a dead-end for sustainable development," he explained.
"Why?" Gao asked immediately.
Su Yuanshan smiled. "Because it creates a chain of problems. You need additional arbitration units to coordinate between cores and caches. It bloats wafer real estate usage, increases power consumption, and worsens heat dissipation.
We can already project future CPU designs based on current data:
Right now, our CPUs have about two million transistors, with manageable power consumption.
Even without radical breakthroughs in design or fabrication, once we move to multi-core architectures, transistor counts will balloon past 30 million.
At that point, a single CPU could easily draw over 100 watts."
"The bigger issue isn't even technical. It's philosophical. True integrated solutions are always more beautiful—and more powerful. But before we get there, I have another idea."
Su Yuanshan turned to face the group. "First, let's squeeze every last drop of performance out of traditional single-core designs."
"Wait," Gao suddenly interrupted.
Su Yuanshan nodded, signaling for him to continue.
Gao's eyes gleamed. "If we're talking about squeezing traditional single-core CPUs, I have an idea—what if we could treat one core as if it were two?"
Su Yuanshan blinked. "Go on."
Get 30% off on my Patreon and enjoy early access to new chapters.
You can also purchase the next 100 chapters of the novel directly from my Patreon page.
Hurry up! The promotion ends on March 2, 2026.
Read 30 Chapters In Advance: patreon.com/Albino1