IBM plans to invest $3 billion in new chips

Release date：2014-10-15   Views：4048

Since the computer age, chips have become smaller and smaller. The Moore's law, described by Gordon Moore, the co-founder of Intel in 1965, still gives a fairly accurate prediction: the number of transistors we put in microchips doubles every 18 to 24 months, continuously improving the speed and efficiency of computers. However, many computer scientists and engineers believe that we will soon be faced with the situation that traditional chip circuits made of silicon will become too small to operate reliably.

So what will happen? No one is sure about this, but chip makers have begun to act to guarantee the future of hardware development. Not long ago, IBM announced a plan to allocate $3 billion over five years for chip research. Although the company's overall R & D spending will remain unchanged, the company's new focus is not only on miniaturizing circuits to 7Nm, but also on replacing silicon chips with unconventional technologies.

"Fundamentally, a 7-nanometer transistor is the size of a large atom, and there are many unknown, uncontrollable quantum effects," said Tom Kant, a computer scientist at Georgia tech. As a result, chip manufacturers cannot guarantee the reliability of their functions. "

Intel is currently able to make 22nm transistors and plans to launch 14nm transistors next year. In general, Moore's law is correct - we've been increasing the number of transistors in a chip for decades. But, according to Cantor, "for a while, the benefits were not great." From 1994 to 1998, the maximum CPU clock speed increased by 300%; between 2007 and 2011, the CPU clock speed increased by only 33%.

"The days of silicon are numbered," Conte predicts. In our situation, we need to step back and rethink how to design a computer. " IBM agrees with Cantor. Several recent announcements of IBM listed several new technologies that can bring breakthroughs in chip development. These new technologies are different in that they can not only make chips smaller, but also make them more efficient and reliable.

A new technology is quantum computing, whose purpose is to improve the computing power of computers. The traditional information unit has only two values of 0 or 1, while the qubit can retain both 0 or 1 values, or both, so that the system can process millions of operations at the same time.

Another new technology is Synaptic Computing. The circuits used in synaptic computing are based on the structures we see in the brain, and the main idea is to allow computers to simulate certain processes that the nervous system is good at processing, such as pattern detection.

Nanophotonics (also known as Silicon Photonics) uses light pulses to process information, rather than electrical signals. In the announcement, IBM said: "we hope nano photonics can provide a superhighway for the rapid transfer of large amounts of information between computer chips on the server."

In addition to silicon, the current chip structure is likely to remain the same. Carbon nanotubes are monoatomic carbon sheets rolled into tubes, which are reported to run 10 times faster than silicon and can easily replace transistor materials.

However, none of these technologies has been tested enough. In addition, some experts have strong doubts about the idea that silicon will be eliminated. "I don't even bet a dollar on quantum stuff." "This kind of thing sounds so unrealistic," says srini davadas, a computer scientist at the Massachusetts Institute of technology In terms of feasibility, carbon nanotubes and nano photonics will not compete with silicon in the near future, he believes. Once we reach 7 nanometers, the miniaturization of transistors is likely to slow down considerably. But there is still a lot of room for innovation in the use of existing materials. "Why don't we develop an alternative silicon material that works?" he asked

Davadas also pointed out that IBM's commitment of $3 billion is nothing compared to what companies like IBM and Intel have invested in silicon innovation research. He believes that as silicon transistors become smaller and smaller, people will be eager to see other technologies bring about a "post silicon era", which will make IBM's announcement more practical.

At the end of the day, no matter how promising other technologies are, one thing is clear - silicon will still be used in this field for at least a few years to come. "Silicon is the current protagonist." "There is no other material that can compete with it," he said It expert network

---

Last：Apple*s double happiness will boost the supply chain

Next：Focusing on mobile and wireless products exhibition, intelligent hardware and wireless charging become absolute protagonists