You are here

GPU-enhanced HPC Design and Optimization Platform for NaNoOptical Components

Award Information
Agency: Department of Energy
Branch: N/A
Contract: DE-SC0018458
Agency Tracking Number: 243756
Amount: $1,494,282.00
Phase: Phase II
Program: SBIR
Solicitation Topic Code: 08a
Solicitation Number: DE-FOA-0001975
Solicitation Year: 2019
Award Year: 2019
Award Start Date (Proposal Award Date): 2019-05-28
Award End Date (Contract End Date): 2021-05-27
Small Business Information
2904 Westcorp Boulevard Suite 210
Huntsville, AL 35805-6410
United States
DUNS: 832864370
HUBZone Owned: No
Woman Owned: No
Socially and Economically Disadvantaged: No
Principal Investigator
 Jeremiah Brown
 (256) 319-2026
Business Contact
 Jeremiah Brown
Phone: (256) 319-2026
Research Institution

There is a fundamental need for a high-performance computing approach to computational electromagnetics modeling tools suitable for designing and optimizing high-end nanostructures and photonics devices. Complex geometries and material responses to incident electromagnetic fields require a significant level of computational resources, especially when realistic manufacturing tolerances are incorporated into the overall design process to ensure good as-fabricated performance. Photonics, semiconductor, and telecommunications manufacturers will greatly benefit from a high performance computing design-for-manufacturing electromagnetic simulation platform capable of optimizing geometries while fully accounting for tolerances. This work will develop a high performance computing platform incorporating finite-difference time- domain algorithms capable of designing, simulating, and optimizing the types of devices needed to support new photonics and nanostructure device concepts. The platform will be designed for massively parallel operation across a large number of distributed nodes with GPU-enabled speed optimization. The Phase I efforts focused on development and validation of FDTD simulation algorithms and of HPC approaches necessary to support parallelized implementations. The FDTD algorithms were developed to incorporate complex material responses, including dispersion and nonlinearities, while the HPC development evaluated different parallelization and memory alignment schemes for optimal computational performance. We will develop a HPC-enabled FDTD CEM package and optimize it for massively parallelized distributed operation across a large number of nodes utilizing CPU and GPU speed enhancements, with automated load-balancing techniques based on machine learning concepts.The simulation package will incorporate global optimization algorithms that account for manufacturing tolerances, allowing the platform to act as a complete design-for-manufacturing package. The design platform will be integrated as a commercial product and will be validated by manufacturing and testing components designed using the software. Commercial Applications and Benefits: This design platform will be of immediate use to semiconductor and telecommunications innovators and to photonics design-for-manufacturing companies. It will also enable IERUS to be a design partner to commercial partners operating in these industries, as well as in biophotonics and Department of Defense applications. The platform will be an enabling technology for developing the next generation components required to facilitate the continued telecommunications growth and will greatly enhance electromagnetic simulation capabilities for customers across a wide spectrum of business sizes and applications.

* Information listed above is at the time of submission. *

US Flag An Official Website of the United States Government