Chapter 92 Starting from MMX
The feedback from Silicon Valley's Xinghai came quickly.
Just as Su Yuanshan had predicted, after Carly and Xi Xiaoding contacted Cyrix, founder Klaus was clearly overwhelmed.
The newly launched CX486DLC series of low-end CPUs allowed Intel 386 users to upgrade to near-486 performance, but the core was still based on a heavily modified 386 and lacked an integrated floating-point unit. Users needed to add an external 80387DX coprocessor.
Thus, even though it was cheap, market prospects were still uncertain — at least, they hadn't received significant local orders.
Don't forget: AMD, the big brother of clones, was still in the way.
Historically, Cyrix only truly gained market recognition after 1995, when the 6x86 and MII chips launched. After making a name for themselves, Cyrix was quickly sold to National Semiconductor.
So when Carly tentatively raised the idea of an acquisition, Klaus didn't outright refuse — he was just very surprised.
Because as the old saying goes, "When something is too strange, there must be a trick."
Klaus said he needed to discuss it with his team carefully.
As the acquisition plan advanced, Yang Yiwen, who had originally planned to stay a month in China, had to cut her vacation short and return to Xinghai to oversee legal and patent transfer matters.
Meanwhile, Qin Si's investment department would spin off into an independent offshore subsidiary of Xinghai, mainly focused on angel investments and patent management.
Another reason for this restructuring was that Chen Haoming had just completed his Ph.D. and was about to officially join Xinghai as Xi Xiaoding's successor.
At Yuanchip, casual office romances were tolerated as long as they didn't involve direct supervisor-subordinate relationships.
For senior management, however, organizational separation was mandatory.
Yuanchip, after all, was a high-tech company — they couldn't run things like a small sales business, always tense about personal matters.
...
After seeing Tang Wenjie and Yang Yiwen off, Su Yuanshan stopped by the construction site again.
Not surprisingly, under the watch of the Japanese construction supervisors, after a full month, the cleanroom was still just at the foundation stage.
Su Yuanshan wasn't in a hurry though — Chen Jing had just successfully sourced a batch of motors via SOYO and was arranging for them to enter the country through Hong Kong.
She told Su Yuanshan bluntly:
"If the normal channels don't work, we might have to use sea routes."
In his previous life, Su Yuanshan had never engaged in this kind of "creative logistics" — he could only say:
"You handle it. I trust you."
That said, he wasn't completely unfamiliar.
For example, the Soviet aircraft carrier Varyag, currently rusting in the Black Sea, would in a few years somehow sneak its way into China — everyone thought she was going to be turned into a floating casino in Macau, but she ended up becoming the Navy's first true flagship, the Liaoning.
...
Back at the office, Su Yuanshan opened a document and began recalling the MMX instruction set.
Instruction sets weren't mysterious.
You could think of them simply as collections of specially optimized commands for certain types of programs.
For example, MMX was designed specifically to accelerate multimedia video tasks — before MMX, playing videos required massive computation.
With MMX, execution efficiency skyrocketed.
As a former Intel engineer, Su Yuanshan had first encountered MMX early on, so he understood it well.
It was, fundamentally, a compromise — designed to accommodate the newly released Windows 95.
Thus, MMX was fully compatible: it didn't require any OS changes to function.
But this compatibility came at a price.
It was bloated, and while it improved 2D video performance, it did almost nothing for 3D.
When AMD later introduced 3DNow!, Intel was forced to scramble out SSE.
As SSE matured and CPU performance improved, MMX faded into irrelevance.
However, for this era — MMX was a devastatingly powerful secret weapon.
Remember, it wouldn't even be publicly proposed until 1996.
Su Yuanshan frowned, spinning a pen between his fingers while staring at the monitor.
Without access to a Pentium CPU design, he couldn't run proper simulations or designs.
All he could do was rely on memory to draft the instruction formats and partial functionalities.
When he eventually took control of Cyrix, he would immediately redesign the CPU core and integrate MMX support.
Actually, he could start designing a new core right now.
But realistically, even if he was a top-tier spy who had spent three years undercover at Intel, it was still delusional to think he could singlehandedly design an x86 core from scratch.
After thinking for a moment, Su Yuanshan picked up the phone and dialed Lin Yonggu's office number.
The line connected.
After a few pleasantries, Su Yuanshan asked:
"President Lin, how's your progress over there?"
Lin Yonggu's voice was calm:
"Not bad. Steady progress."
"Good, then allow me to wish you success in advance," Su Yuanshan said with a pause, smiling faintly. "Haha... President Lin, when will my people be coming back?"
Lin Yonggu froze for a moment, then replied urgently:
"They can't leave yet! We're at a critical stage!"
Su Yuanshan chuckled:
"I didn't mean immediately. We're partners — if we need them back, I'll definitely give you plenty of notice."
Lin Yonggu finally relaxed and laughed:
"I thought your government was pressuring you to pull them out."
Of course he suspected that — sending a whole team over obviously wasn't charity.
If they were being recalled, it usually meant outside pressure.
To be honest, at first, UMC's people had underestimated Yuanchip's engineers.
But after the initial adjustment period, as projects advanced, they realized Yuanchip's engineers weren't amateurs — they were genuinely talented.
Smart, efficient, professional.
Now Yuanchip's team had become so integral to the project that if they left, progress would stall immediately.
"Hehe, by the way, the YX chip my dad showcased during his visit — that's the same proprietary architecture I mentioned before," Su Yuanshan said softly.
"If you need it, we can license it to you."
"I know you guys want to ride the x86 wave too. But with Intel suing AMD and Cyrix now... can your CPUs really dodge the 338 patent?"
Lin Yonggu went silent. After a long pause, he laughed bitterly:
"No. No one can."
"If you get sued, can you handle it? If you can, maybe you can focus on mainland markets. If not..."
Su Yuanshan trailed off meaningfully.
"My personal advice is to embrace the YX architecture."
After another silence, Lin Yonggu finally chuckled:
"We're almost ready for simulations now.
As a CPU architect... I need to at least finish designing one chip to be at peace."
Su Yuanshan sighed softly:
"I understand. Well, best of luck — which means my engineers should be back in about three months?"
"Yes, at most three months."
"Good."
After hanging up, Su Yuanshan exhaled deeply.
Three months.
By then, Huajing's new production lines would be operational, and Yuanchip's architectural plans would also be ready.
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