Advancing computing in a post-Moore's Law world
Connecting state and local government leaders
DARPA plans to use new computer architectures, design and materials to push computing, even as Moore’s Law seems to be reaching its limit.
When the Defense Advanced Research Projects Agency launched the Electronics Resurgence Initiative last year, it aimed to use new computer architectures, design and materials to continue computing advances even as Moore’s Law seems to be reaching its limit.
Last month, DARPA announced the teams that would be leading this research. ERI is divided into three “thrust areas”: architectures, design and materials and integration.
The architecture group will research software-defined hardware and frameworks that can solve specific computing problems more quickly and efficiently by changing their configurations in real-time based on the data they are processing. The group is also researching domain-specific system-on-chip design, which seeks to determine the right amount and type of hardware specialization while making a system as programmable and flexible as possible, DARPA officials said.
The first domain that DSSoC researchers will tackle is software-defined radio, a technology critical to mobile and satellite communications, personal area networks, radar, and electronic warfare.
“These devices must be programmable like general purpose processors, but also capable of crunching a lot of math with low power," said Tom Rondeau, the Microsystems Technology Office program manager leading DSSoC. “It is critical for the DOD to have flexible, adaptable radio systems that are capable of managing and combating a complex signal environment.”
The design group will be working to rethink the semiconductor. Researchers want to take the human out of the picture when it comes to hardware design and use machine learning to improve the performance of multiple computer components.
The final group, materials and integration, will be focusing on computer memory and, more specifically, the “memory bottleneck” caused by the time and power required to access memory. It will explore integrating memory and logic into a 3D monolithic system-on-a-chip stack.
The winning teams are made up of universities, national labs and private companies like NVIDIA and Intel.
ERI is a five-year effort that will pour up to $1.5 billion into funding “to jumpstart innovation in the electronics industry,” DARPA officials said.
NEXT STORY: Bringing drones into a smart city network