A Massive Datacenter On Single Chip
Researchers have been designing datacenter-on-chip (DoC) technology for faster and more energy-efficient data processing and better performance for big data applications.
The project has received a grant from the US Army Research Office to develop a novel computing platform for emerging big data applications. Datacenter-on-chip technology has been developed by a group at Washington State University and Carnegie Mellon University.
The research team has put the equivalent of a huge data center, which uses enormous amounts of energy in large facilities to crunch data for companies like Google or Amazon, on a single computer chip. The chip includes thousands of processors, or cores.
The US Army is interested in such many core platforms for large real-time battle simulations and for battle information management software. Many military applications such as electronics for war fighters and unmanned aerial vehicles also require both high performance computing and low power consumption.
According to Washington Sate University, the researchers are working on a technology that is specifically optimised for deep learning.
Since deep learning techniques are very data intensive, the researchers will use graphics processing units (GPUs) on the chip to speed up processing in addition to central processing units (CPUs), creating what is known as a heterogeneous platform.
The researchers evaluate that the main challenge is to integrate CPUs and GPUs into a single chip with an efficient communication network between them.
They are considering hybrid communication networks consisting of both wired and wireless connections to address this important challenge. Amazon is also underway with the design of the second generation of its data centre chips.
You Might Also Read:
Business Migration To The Cloud: