2024-06-30 00:03:22
Specialists from the College of California, Santa Barbara (USA) have discovered a solution to considerably enhance the power effectivity of computer systems. The examine was revealed within the journal Nature, Day.Az studies with regards to Lenta.ru.
In response to scientists, computer systems concerned in advanced calculations considerably exceed the capabilities of the human mind. Nonetheless, their power effectivity is 4 orders of magnitude or 10 thousand occasions worse. Engineers have proposed creating computer systems primarily based on neuromorphic computing in order that they eat much less power.
The article states that within the human mind, information processing happens in parallel with the participation of many neurons with low power consumption. Scientists have proposed utilizing tunnel field-effect transistors (TFET) primarily based on transition metallic dichalcogenides (TMD) within the creation of machines.
The system proposed by specialists begins solely when there may be enter information for processing. In addition they observed that in supercomputers, many of the power is misplaced attributable to leakage when a selected circuit is within the off state. The researchers believed that there can be much less leakage if tunnel transistors with decrease present had been used.
“As soon as this platform is carried out, it may carry the facility consumption of chips to 2 orders of magnitude in comparison with the human mind,” the examine authors stated. Thus, the mind might be solely 100 occasions extra environment friendly than a pc, not 10 thousand.