Rebirth: Starting from Lighting Up the Tech Tree

Chapter 10 Compiler Maniac

Fang Ze was thinner than Zuo Cheng had imagined.

He was over 1.8 meters tall and probably weighed less than 130 pounds. He had prominent cheekbones and slightly sunken eyes, and his whole appearance seemed to have been drained of moisture by coding. But his eyes were very bright—the kind of brightness that only someone who has been thinking intensely for a long time possesses. When quiet, they were like two light bulbs that had gone out, but once the topic turned to technology, they would light up instantly.

Thursday evening, at the barbecue stall near the school gate. Zhang Lei treated everyone, ordering a table full of skewers.

Fang Ze sat opposite Zuo Cheng, barely touching the beer in front of him, but he ate three grilled buns dipped in chili sauce.

"Zhang Lei said you guys set up a studio and take on technical projects on the forum?" Fang Ze's voice was not loud, and his speech was slow, as if he had rehearsed each sentence in his mind before speaking.

"Yes, it's just a small-scale project." Zuo Cheng didn't rush to promote it, but instead started a conversation, "I saw your microkernel project on the forum, and there's something I'd like to ask—your task scheduler uses priority inheritance to solve the inversion problem, but I noticed that you didn't use the conventional priority ceiling protocol, but instead designed your own dynamic inheritance chain. Why?"

Fang Ze's eyes lit up.

"The priority ceiling protocol is too brute-force." He put down his steamed bun and began gesturing on the table with his fingers. "It directly raises the priority of low-priority tasks to the highest level. While it does prevent inversion, it causes unnecessary preemption delays—while high-priority tasks are waiting, all other medium-priority tasks are suppressed, causing a direct drop in system throughput."

"So you use dynamic inheritance chains for precise control."

"Yes. Only the priority on the direct dependency chain is inherited, without affecting unrelated tasks. This solves the inversion problem without significant throughput loss." Fang Ze paused here, glanced at Zuo Cheng, and said, "You can see this design difference, which shows you read code very carefully. Are you an embedded systems engineer?"

"Not exactly signal processing. But I've been paying attention to embedded systems."

"Signal processing?" Fang Ze thought for a moment. "Then you should know how demanding real-time signal acquisition is for the operating system's scheduling—if the interrupt response delay exceeds five microseconds, the sampled data may be dropped. My microkernel's interrupt response can currently be within three microseconds, but I think there's still room for improvement."

"Three microseconds is already very impressive." Zuo Cheng was telling the truth. The interrupt response time of commercial real-time operating systems is usually five to ten microseconds. Fang Ze's microkernel, which he developed by hand, achieved three microseconds, which is already a professional level.

The two men talked about everything from the scheduler to memory management, and then from memory management to the optimization of the interrupt vector table. Zhang Lei and Liu Wei couldn't get a word in edgewise and could only eat their skewers in silence, occasionally exchanging a knowing look that they didn't understand but thought was impressive.

After talking for nearly forty minutes, Zuo Cheng had a clear understanding of Fang Ze's technical skills—this man's depth of understanding of embedded systems was already comparable to many engineers with three to five years of experience. Moreover, he wasn't just a bookish nerd obsessed with technology; he had excellent taste in architectural design and knew where to refine and where to make trade-offs.

The technological radar was quietly scanning Zuo Cheng's consciousness.

[Target: Fang Ze]

Mastering Technology: Embedded Real-Time Operating System Design (Proficient Level)

[Copying costs: 3 points]

Proficient level. One level higher than Yu Ying's "Skilled" level, but one level lower than the "Expert" level from Starry Sky Technology. It costs three points, and he currently has five, so he can afford to spend them.

But Zuo Cheng didn't rush to replicate it.

It's not that I'm reluctant to give up points, it's that it's unnecessary.

Fang Ze's embedded systems skills are indeed strong, but Zuo Cheng's main focus right now is on the Blue Bay Communications project, specifically communication algorithms, which are not relevant to embedded systems for the time being. With three points spent, the Technology Radar will enter a 72-hour cooldown period. It would be awkward if they encountered a more valuable subject for replication in the communications field during this time.

Moreover, what he wanted more was not Fang Ze's skills, but Fang Ze himself.

Technology can be copied, but a teammate who can independently design the operating system microkernel cannot be copied.

"Fang Ze." Zuo Cheng put down his chopsticks, his tone shifting from technical discussion to business. "To be frank—I'm not trying to get you to take orders and make money."

Fang Ze raised an eyebrow slightly.

"Studio 402 is indeed providing campus-level technical services at this stage, but this is not the end. I have a horizontal research project with Blue Bay Communications, which focuses on 5G channel estimation algorithms. This algorithm ultimately needs to be implemented in a real base station environment, and the embedded platform of the base station has extremely high requirements for real-time performance—the signal processing module must run on a customized embedded system, and interrupt latency, task scheduling, and memory management are all hard specifications."

Fang Ze's gaze changed, switching from casual conversation to attentive listening.

"you mean--"

"I don't need you to write code right now," Zuo Cheng said, looking at him. "What I need is someone who can communicate with me at the embedded level. Once the project progresses to the algorithm deployment stage, I'll need someone to port the algorithm to the embedded platform. That's not something just anyone can do; it requires someone who truly understands the underlying mechanisms."

Fang Ze remained silent for more than ten seconds, his fingers unconsciously twirling the beer bottle.

"You're quite an interesting person," he said. "When others come to me, they immediately talk about money, projects, and what kind of work they can do. You're the first person who talks to me about technology for forty minutes first, and then tells me, 'We don't need you to write code right now.'"

"Because you're not someone who lacks work," Zuo Cheng smiled. "What you lack is a direction worth your full commitment."

Fang Ze stared at him for two seconds, then smiled.

That was the first time he smiled that night, and when he smiled, his thin face didn't seem so cold anymore.

"Okay. Let's not talk about joining yet. Send me the technical details of your project. If the direction is indeed interesting, I can participate as a consultant first, and then we can discuss it further when it needs to be implemented in embedded systems."

Zuo Cheng nodded: "I'll compile a technical document and send it to you tomorrow."

The two clinked their beer bottles. Zhang Lei let out a long sigh of relief and whispered to Liu Wei, "It's done, it's done. Cheng Ge's ability to make empty promises is amazing."

Zuo Cheng, with his sharp ears, heard it but ignored him.

On the way back from the barbecue stall, the night breeze, carrying the chill of late September, swept through the sycamore-lined avenue. Zuo Cheng lagged a few steps behind and opened the system panel.

Three emerald green leaves, five points, and the main quest chain countdown is 21 days.

The outline of the fourth blade, "Embedded System Design," was still light gray, but he had a premonition that if the collaboration with Fang Ze deepened, the system would eventually generate new tasks in this direction.

Not urgent.

The growth of the technology tree requires nutrients, and nutrients are not just points and tasks—connections, teams, and resources are all part of the root system.

The phone rang.

Yu Ying sent a long message. Zuo Cheng opened it and his pupils suddenly contracted.

"Zuo Cheng, I conducted another in-depth analysis of the anomalous data under low signal-to-noise ratio conditions today. The conclusion is basically confirmed—it's not data error, but a nonlinear coupling effect between your adaptive tracking mechanism and my high-dimensional decomposition module. This effect is not obvious under normal signal-to-noise ratio conditions, but it is activated in extremely low signal-to-noise ratio environments, equivalent to the two modules working together to perform 'spontaneous noise reduction.' I checked the literature from the past five years and found no similar reports. Zuo Cheng, if this can be theoretically proven, it's not just a research project-level achievement—it's a paper-level achievement. And not just an ordinary paper, but the kind that can be published in top journals."

Zuo Cheng stood under the sycamore tree, the streetlights casting dappled shadows on his face.

Top-tier journal.

The Blue Bay Communications project he was working on was already quite significant, but if it could also lead to a top-tier journal article during its development—

Then his graduation thesis is no longer a matter of "provincial excellence".

That's something that directly rewrites a resume.

He took a deep breath and replied to Yu Ying with two words:

"Dig deeper."