Chapter 47 Cache

The scale of Fuxing Base is still small now. When the scale expands in the future, the base cannot be operated without purchasing hundreds or thousands of units.

In particular, Han Yang also plans to dig the underground base after the computing power is upgraded again.

The Fuxing Mall on the surface is just a cover. For safety, Han Yang decided at the beginning that he must develop underground in the future.

However, Han Yang certainly cannot find a large number of workers and mobilize a large number of heavy equipment to dig the underground base.

He still has to use his own computing power to control the robots to dig.

By then, the requirements for computing power will be higher. And after the base is built, in order to operate more equipment and conduct more scientific research, the number of robots required will not be small.

Imagine the scene of controlling thousands of robots busy in the base with greater computing power in the future, and Han Yang's heart is full of expectations.

But at this stage, we can only look forward to it. If we can't develop more advanced chips and improve computing power, everything is just a mirage.

Time passes slowly. Half a month after the first trial production, on this day, the robot still moved a box of materials piled in the corner of the laboratory to the source of the production line, unpacked it, and poured the materials into the feed port as usual.

The fixed robot worked with the machine to start processing this batch of materials. The ten mobile multifunctional robots shuttled back and forth in the production line like conscientious workers.

Feeding, operation, sorting, integration, and discharging, everything was in order.

Han Yang was the only one in the entire huge laboratory. At this moment, he was still listening to a lecturer from the Department of Mathematics in front of the computer. Behind him, everything was handed over to the machines and robots to operate, just like a fully automated assembly line.

When this batch of raw materials was transformed into a specific type of cable and came out of the discharge port, and then assembled into bundles by the robot and sent to the special testing equipment for inspection, the numbers on the instrument came into Han Yang's eyes, which immediately cheered Han Yang up.

After half a month of repeated modifications, debugging, and attempts, the first batch of cables that met Han Yang's needs were finally produced.

A total of 100 cables were produced in this first batch. They are 12.5 cm long, with a small connector at both ends, and look very delicate.

On the other side, another device was also roaring, producing some things as small as rice grains, which were also sent to the inspection equipment.

These things are all transistors.

Cables are equivalent to circuits, and circuit boards are equivalent to silicon wafers. Silicon wafers, circuits and transistors are the three most important things to build a CPU.

In regular CPU production, transistors and circuits are on silicon wafers, directly photolithographically produced with light, and there is no need for such trouble.

At this moment, Han Yang can only achieve the replacement of photolithography technology in this way.

At this moment, the circuit boards that meet Han Yang's needs, as well as the first batch of cables and transistors, are ready.

As soon as Han Yang thought about it, a multifunctional mobile robot carefully picked up these materials and placed them on the work platform.

The fixed robot on the work platform immediately welcomed Han Yang's computing power infusion.

Under Han Yang's direct operation, the new generation of CPUs finally officially began to be manufactured.

During this period of long-term learning, Han Yang has gradually conceived the architecture and manufacturing methods of the new generation of CPUs in his mind.

Compared with the previous generation of CPUs, the logical structure of this new generation of CPUs is more complex. Han Yang added multiple self-created instruction sets to it and adopted the pipeline design mode for the first time.

Pipeline circuits greatly improve efficiency by allowing multiple instructions to perform different tasks at different locations on the circuit at the same time.

This is a very mature technology. In modern society, even the oldest CPUs have adopted this technology in large quantities. But for the circuit board CPU that Han Yang rubbed out by hand, this technology is still the first time to be used.

According to theoretical predictions, this technology will greatly improve the performance of the CPU he manufactured.

In addition, Han Yang also added a high-speed cache of up to 4KB to this CPU for the first time.

Modern CPUs generally have three levels of cache to maximize performance. But for Han Yang, making a cache structure is already the limit of existing technology.

The addition of many new designs makes this CPU architecture much more complicated than before, and correspondingly, the manufacturing difficulty has also increased a lot.

Although Han Yang had designed this architecture in advance, he knew that theory was only theory after all.

In the specific manufacturing, there would be many new problems that he had not discovered before.

Solving a problem is the only feasible solution for Han Yang at this moment.

As expected, the first batch of 500 transistors and 100 cables had not been used up before the problems emerged.

The welding precision was not enough.

The cables and transistors were too small. When they were fixed on the circuit board by welding, the welding ports would affect each other, causing a short circuit.

This was a problem that had never occurred before.

Han Yang spent a lot of time thinking about this problem. In the end, Han Yang spent a lot of money to buy a high-precision welding gun and purchased special solder to solve this problem.

For this reason, the first batch of transistors, cables, and circuit boards were all scrapped and turned into garbage.

Fortunately, the second batch of transistors and cables had been made, so Han Yang didn't have to wait any longer.

After solving this problem, Han Yang continued to manufacture.

Then soon, the second problem appeared again.

Han Yang found that there was a bug in his original design. The two modules in the CPU were too far apart and were located on two different circuit boards. They needed to be connected by a cable with a length of 62 cm.

The cables manufactured by existing technology could not achieve the required accuracy and speed.

There was no way, Han Yang could only adjust the architecture, build the two modules on the same circuit board, and move the modules originally located here to the rest of the circuit boards.

But this would affect the overall structure, and the communication reliability, communication delay, etc. between all modules would have to be reconsidered.

Fortunately, this was just an adjustment to the overall structure, not involving details, and it was not too troublesome. Han Yang spent five days to adjust all of this.

After that, the manufacturing work continued.

After the third day, Han Yang stopped working again, and started researching and thinking again. After a two-day break, he continued working, and then stopped again after four days. After that, he stopped again after another three days...