The technical staff at SOFAR Acoustics provides a synergy of experience in acoustics, signal processing, sonar technology, seismology, and information technology. We offer a variety of applied research and development services, from acoustic modeling, simulation, and signal processing, through system and sensor development, to distributed cloud computing. We provide services related to following disciplines:
■ Signal Processing
■ Embedded Linux Systems
■ Distributed Sensor Networks
■ Cloud Computing
GPU Acceleraiton of AUV Processing
There exists a need for innovative, cost-effective, parallel architectures for autonomous acoustic processing to significantly increase the processing capability for new autonomous underwater sensing systems.
SOFAR Acoustics is conducting an effort to deliver a cost-effective, low-power, highly accelerated execution environment for the underwater acoustic signal processing on Autonomous Underwater Vehicles (AUVs), addressing diverse operational constraints given by space, power consumption, numerical accuracy, and access to communication infrastructure.
Most GPU architectures, such as the top-performing Nvidia Kepler K-40, can deliver massive amounts of parallel computation through hundreds of arithmetic-logic unit lightweight cores, but require approximately 300 Watts to operate and are therefore not viable options for these environments.
SOFAR Acoustics can provide a cost-effective COTS hardware/software processing acceleration solution capable of perform within a single-digit Watt power-budget, while at the same time providing speedup of at least five times over the current single threaded processing.
Copyright © 2010-2014 SOFAR Acoustics - All rights reserved.