In an attempt to enhance understanding of neurological diseases, scientists have made an energy-efficient and fast simulation of rat brain by employing AI (Artificial Intelligence) compute platform from Nvidia—the computer processor maker.
Developing more efficient and faster simulators can elevate the level of understanding the function of the brain and verify how injury to a specific structure in neurons can result in discrepancies in the function of the brain. For making the simulator, the scientists employed computer hardware developed for 3D games, as per the study posted in the Frontiers in Neuroscience journal.
The research displayed that a single GPU (Graphics Processing Unit) was capable of attaining processing speeds of almost 10% quicker than what is presently possible employing either the SpiNNaker neuromorphic system (a custom-developed device) or a supercomputer.
The team was also capable of achieving energy savings of 10 times in comparison to either a supercomputer or the SpiNNaker simulations.
On a related note, computer devices of systems such as a neural firing or traffic flow of the city in the brain tend to employ up a huge amount of memory. But a new method with quantum simulators can considerably slash that memory consumption by taking quantum advancement to time. The only expanse is a reduced record of history.
The recommendation arrives from scientists Thomas Elliott and Mile Gu in Singapore, who define their bid in a paper posted in npj Quantum Information. Gu operates at the NTU (Nanyang Technological University) and the Centre for Quantum Technologies in Singapore, and Elliott works at NTU.
To carry out replication, a classical device must slash time into separate actions. Gu summarizes similarity with a very old method of calculating time: the hourglass. “Focus on an hourglass and one can look at the separate grains of sand dripping down individually. It is a granular course,” claims Gu.