Chapter 86: 20-Nanometre Chip
The drill method Tom adopted was to intentionally delay issuing commands, so that both fleets would have a communication delay of about 100 seconds with him, but one side would not have combat AI assistance, while the other side would.
Moreover, the combat AI was deployed near the battlefield, with no delay.
As soon as the two sides engaged in combat, Tom noticed something was wrong.
Is this artificial intelligence? This must be artificial idiocy, right?
Clearly, a warship in a melee should push forward to provide fire suppression for its allies, creating conditions for them to launch interstellar missiles.
However, during this 100-second communication delay, before he could issue the next command, the combat AI preemptively controlled the warship to retreat directly, fleeing in a panic.
It did indeed dodge an enemy interstellar missile, but it directly tore apart its own defense line, disrupting its own battle plan, which immediately led to the enemy seizing the opportunity to unleash a volley, firing millions of rounds, causing heavy losses to its side.
Another time, this side should have launched an attack, but under the control of the combat AI, their warships remained motionless, ultimately causing them to miss the opportune moment.
These were relatively major errors. The other, more minor errors were even more numerous, countless.
Ultimately, in this drill, the side without combat AI assistance, with Tom enduring a communication delay of about 100 seconds, achieved a complete victory, "annihilating" the side with combat AI assistance.
"More than ten years, over 100 million tons of fuel consumed, trillions of bullets consumed, and this is the result?"
Tom was simply unable to accept it.
"That shouldn’t be right? How could this happen?"
Tom recalled the cargo ship carrying the supercomputer from the battlefield, retrieved the combat AI’s operational data, and dispatched thousands of Clones to analyze its code execution and decision-making logic.
After this analysis, Tom gradually discovered the problem.
He seemed to have misunderstood this set of combat AI...
Because he underestimated the difficulty of real-time control of warships on the battlefield, as well as the amount of data that needed to be analyzed and computed.
It wasn’t that this AI was idiotic; rather, at the beginning of the battle, the data volume was too large, the computational load was too great, which resulted in the supercomputer being overloaded and unable to keep up, leading to a large number of delays, computational errors, and logical errors, ultimately causing this situation.
Tom breathed a sigh of relief.
"Fortunately, fortunately it wasn’t a problem with the combat AI; these ten-plus years of investment don’t seem to have been in vain..."
Once the problem was found, the rest was easy to discuss.
Since the supercomputer’s computing power was insufficient, then develop a supercomputer with greater computing power!
At this moment, with parallel algorithms and structures having almost reached their limits, there was only one way left to create a supercomputer with greater computing power.
Develop more advanced chips!
At this stage, Tom has mastered the mass manufacturing technology for 45-nanometer process chips.
According to the technical roadmap already practiced in the human world, the next generation for Tom should be to develop 28-nanometer process chips.
However, combining years of continuous research and technological advancements, Tom decided to bypass 28-nanometer chips and directly research the next generation: 20-nanometer process chips!
Once the 20-nanometer chip is successfully developed, the number of transistors per single chip will skyrocket from approximately 1 billion to 4 billion.
The overall performance of a single chip will also increase by about 60% to 100%.
But this is not the most important thing; the most important thing is that higher-process chips can adapt to more advanced parallel algorithms and supporting hardware. In a single supercomputer, Tom can use more chips for parallel computing, and the computing power of one supercomputer can be increased by about 40 times on the existing basis!
The prospects are so good, but directly upgrading from 45-nanometer to 20-nanometer, bypassing 28-nanometer, obviously means huge obstacles and extremely high difficulty.
At the Computing Technology Research Base, Tom once again dispatched numerous Clones to begin research on more advanced chips, building upon existing research.
After a short period of preliminary research, the technical difficulties he needed to overcome were categorized into several major types and presented before Tom.
"First, the deep ultraviolet light source I am currently using has too long a wavelength, and its resolution has reached its limit. Diffraction effects cause blurred patterns, making it impossible to engrave smaller transistor structures.
Secondly, traditional transistors at this size exhibit significant short-channel effects, making power consumption and performance difficult to control.
Also, the physical properties of traditional silicon gate dielectrics do not meet the requirements; this also needs to be improved..."
Tom divided the Clones into several large teams, each tackling different aspects of the problems.
During the continuous research, breakthroughs first appeared in the light source.
Tom developed an extreme ultraviolet light source with a shorter wavelength, reducing the wavelength of light to only about 13.5 nanometers, thereby greatly improving resolution.
In addition, Tom also changed the previous process, converting the previous single exposure to multiple exposures, which allowed for a significant improvement in processing accuracy by increasing the process flow.
After that, Tom also found better materials, using metal gates to replace silicon dioxide gates, achieving another technological breakthrough.
Six months had passed by this point. Most of the other technical problems had been overcome by Tom, but only the most difficult and important technical obstacle still stood before him.
Short-channel effect.
When the length of a transistor is reduced below a certain extent, the electric field distribution and carrier behavior will change due to physical laws, and this will lead to a series of problems such as a decrease in the transistor’s threshold voltage and a reduction in the drain-induced barrier, which in turn severely affects chip performance and reliability.
To solve this problem, Tom had tried thousands of solutions, including optimizing transistor distribution and design, improving material performance, and refining production processes, but all were ultimately proven ineffective.
"Other problems have been solved, but this one cannot be solved, this is truly..."
Tom sighed.
At this moment, the time and resources consumed by the 20-nanometer chip research had already exceeded expectations, affecting Tom’s overall scientific research plan.
But there was no way; giving up halfway was impossible. The 20-nanometer chip was too important to Tom; not only would the combat AI use it, but other AIs, including factory intelligent equipment, spacecraft intelligent control, and even research in fundamental physics, mathematics, chemistry, and engineering, all depended on supercomputers with greater computing power.
On this day, Tom continued to control tens of thousands of Clones to conduct relevant experiments, and additionally mobilized thousands of brainpower specifically to ponder this problem.
It was under these circumstances that an idea suddenly generated in the mind of a certain Clone, transmitting into Tom’s consciousness.
"Since planar transistors lead to short-channel effects, then... can we change the transistor from a two-dimensional structure to a three-dimensional one?"