Chapter 155: Memory Solution

Although Civilization was a turn-based strategy game and didn't require the kind of raw processing power that real-time games did, that didn't mean it was easy on your hardware.

In fact, it was relatively light on processing power. The CPU barely broke a sweat, and unless the graphics were particularly ambitious, the graphics chip didn't have much to do either. After all, this wasn't The King of Fighters, with dozens of high-speed sprites flying across the screen.

But there was one thing Civilization could chew through surprisingly fast: memory.

An enormous number of map tiles, units, and cities, along with the state of every object and a snapshot of every turn, all consumed massive amounts of memory. That was the real headache.

After making a rough estimate of how much memory the game would need, Ethan felt a chill run down his spine.

"That old saying really wasn't wrong."

In the short term, you run out of chips; in the long term, you run out of energy; but given enough time, you'll always run out of storage.

Whether it was RAM or long-term storage, memory had been a constraint from the very beginning, binding him like an unbreakable chain.

The Nintendo Famicom had just 2 KB of RAM and 2 KB of VRAM, while Super Mario Bros. took up only about 40 KB. Compared with GAMENOVA's games, which routinely exceeded 1 MB, the difference was almost impossible to imagine.

"I do have a more advanced memory design..."

Now that Ethan had a solid grasp of semiconductor technology, creating faster, higher-capacity memory wasn't particularly difficult.

The problem was... "It's ridiculously expensive."

Memory was worth its weight in gold in every era.

A single 1 Mb memory chip already cost more than 30 US dollars to manufacture. If he wanted to equip the NovaXeno with 1 MB of RAM, the memory alone would cost over 300 dollars, and for a home game console, that was simply absurd.

"The Sega Mega Drive's specs..."

72 KB RAM + 64 KB VRAM + 8 KB Audio RAM.

The console, released back in 1988, was every bit as overwhelming against its competitors as the NovaBox had been among 8-bit systems. It was a complete generational leap.

"Hey..."

Sam suddenly interrupted Ethan's muttering. He had been listening to Ethan mumble to himself for nearly half an hour and had finally pieced together what was bothering him.

"Why not just leave an expansion slot for extra RAM? Let players decide how much memory they want."

"...?!"

Ethan shot upright.

"Holy crap... you're right!"

There was no reason to build the NovaXeno as a monster that completely outclassed Sega's upcoming Mega Drive. What Ethan needed was a console that could evolve.

They could combine SGI's technology, NeXT's architecture, and Dell's experience with modular hardware to create something closer to a custom-built PC than a conventional game console. Give the system expansion slots for graphics cards and add 30-pin SIMM slots for additional memory, and the NovaXeno could be upgraded as technology advanced.

By the time Sega launched its next-generation console, compatible NovaXeno upgrades could already be on the market. Instead of becoming obsolete, the console could simply be upgraded to keep pace with newer hardware.

Ensure your favorite authors get the support they deserve. Read this novel on the original website.

That also meant they didn't need to cram huge amounts of memory into the base model. With careful optimization, even 512 KB, or perhaps just 256 KB would be enough.

And Ethan already had something even better up his sleeve: a way to make that limited memory perform several times better than it should.

"Tell Gerald to update the hardware design. I'm going to see Vivian."

...

"An improved memory architecture? What exactly do you mean?"

Vivian stared at Ethan in confusion. She already knew he had been aggressively recruiting integrated circuit design teams across Silicon Valley, with some of them already making the transition to MicroNova.

She had assumed he simply wanted to build his own semiconductor fabrication plant to reduce memory production costs. But after hearing what Ethan was saying now, she realized that wasn't his plan at all.

"That's right, I designed this a while ago, and I call it the SDRAM architecture."

"SDRAM?"

Vivian tilted her head.

Graphics hardware was her specialty, but she also had a degree in electrical engineering and understood memory technology far better than the average person, so hearing a name she had never encountered immediately caught her attention.

"Is it some kind of hybrid between DRAM and SRAM, like a form of high-speed memory?"

When people talked about high-speed RAM, SRAM was the obvious comparison, but its extremely high manufacturing cost and much lower capacity made it impractical for general-purpose memory and limited it mostly to CPU cache.

"No, the 'S' stands for Sync, as in synchronous."

Ethan smiled as he explained.

Memory technology in the 1980s was nothing like the DDR memory that would become standard decades later, but the potential profits hidden in this technology were enough to make every semiconductor company in the world lose sleep.

Unfortunately, there was a problem. DDR manufacturing processes pushed directly against the technological limits of the era, meaning that even with complete architectural blueprints, mass production was still impossible. But DDR's predecessor, SDRAM, was something they could actually build.

"With this design, we add a synchronized clock signal to the memory, allowing it to operate in sync with the CPU."

It sounded like such a simple change, but the impact could be enormous because synchronization could increase memory performance severalfold by eliminating one of the biggest inefficiencies in traditional memory systems, something this era could barely imagine.

"A synchronized clock?" Vivian asked. "What does that actually accomplish?"

"It makes the CPU faster."

"...Huh?"

Smiling, Ethan pointed toward a pen lying on a nearby desk.

"Could you hand me that pen?"

Although puzzled, Vivian obediently walked over and picked it up.

Ethan raised the pen.

"Let's say this pen represents traditional asynchronous memory, I'm the CPU, and you're the memory; I issue you a command telling you to retrieve data from a specific address, you receive the request and start looking for it, while I have no choice but to stand here and wait until you bring the data back before I can continue."

Vivian's eyes immediately lit up as the idea clicked.

"I get it! Synchronous memory eliminates that wasted waiting time, so the CPU can work on something else while the memory handles the request."

"Exactly."

With traditional asynchronous memory, the processor had to wait for the memory to respond before moving on to the next instruction, and although a single wait might last only 100 nanoseconds, billions of such delays could add up to an enormous amount of wasted processing power, leaving the CPU sitting idle when it could have been doing useful work.

A synchronized clock changed that, it allowed the CPU and memory to coordinate their operations far more efficiently, and the advantage went beyond simply reducing wait times because memory could also begin preparing the next request while the previous one was still being processed.

In other words, while the first block of data was being read, the memory could already receive the address for the second request, creating a continuous stream of data instead of forcing the CPU to stop and wait after every operation.

"Genius!" Vivian exclaimed. "This is next-generation technology!"

"Sam is already putting together a patent pool."

Ethan placed a hand on her shoulder, his expression turning serious.

"I want MicroNova to own the core patents for this new memory architecture."

Those foundational patents absolutely had to remain in their hands, because once MicroNova established a patent pool and entered into cross-licensing agreements with semiconductor giants such as Micron and others... the technology could become an almost inexhaustible source of revenue for them.

It would literally be a money printing machine, and the potential wouldn't end with the first generation.

From DDR1 all the way through DDR5, the underlying principle would remain remarkably consistent, because despite all the advances that followed, the entire DDR era would ultimately trace its foundations back to one simple idea Ethan had just introduced: synchronous memory architecture.

-----------------------

Guys if you liked the chapter just leave some comments and reviews, it's fun to read them.

And if you want some extra chapters, be sure to check out my Pa-tron page, it's 60 chapters ahead and the chapters are also long so you can check it out.

p at re on . co m / NovaScribe

(Just remove all spaces and then search)