GSI Technology FRACTALS Project Partners with SHREC on Space-Related Computing Systems

SUNNYVALE, Calif., July 28, 2020 — GSI Technology, Inc., supplier of high-performance reminiscence options for the networking, telecommunications, and army markets, introduced that SHREC (NSF Center for Space, High-Performance, and Resilient Computing) has partnered with the GSI Technology venture FRACTALS [Fault tolerant and Resilient Associative CompuTing for Artificial inteLligence in Space].

The aim of FRACTALS is to create modular and cost-effective computing methods for all space-related efforts, from ground-based high-performance computing (HPC) knowledge facilities to deep-solar-system exploration missions. GSI Technology’s synthetic intelligence (AI) group is dedicated to the open nature of AI. FRACTALS welcomes collaborators, like SHREC, to hitch this effort bringing novel computing architectures to mission-critical methods in area.

Lee-Lean Shu, Chairman and Chief Executive Officer of GSI Technology, commented, “In our partnership with SHREC, GSI gains access to thought leaders and researchers to help us advance the development and commercialization of our Associative Processing Unit (APU) processor. SHREC addresses research challenges facing three domains of mission-critical computing – space computing, high-performance computing, and resilient computing. Through its Industry-University Cooperative Research Centers (IUCRC) programs, the organization develops and supports relationships between industry innovators, world-class academic teams, and government leaders. GSI will leverage these resources to advance the APU’s performance and expand our network in this sector.”

“Two space-related trends that benefit Gemini, our APU processor, are the increase in satellite launches and far-away space mission deployments,” continued Mr. Shu. “For advanced HPC systems on Earth processing the incoming data from satellites, our Gemini delivers significant AI processing advantages in speed and power. Over time, as more processing ability resides in space, processors will need high performance and the ability to tolerate harsh radiation. Radiation tolerant Gemini can provide onboard AI in severe conditions and process the massive amounts of data required with a lower power requirement.”

Deep studying can devour monumental quantities of knowledge and distill all of it into compact machine studying fashions. On Earth, educated fashions may be analyzed in superior HPC methods to carry out numerous features like object recognition, language understanding, predictive analytics, and even advanced decision-making. In area, these processors can use educated fashions to handle large quantities of sensor knowledge to interpret intelligence, predict mission-critical occasions, facilitate human-computer interplay, and even empower local-vehicle autonomy.

As area missions transfer past close to Earth satellites to the Moon, Mars and different planets and probably past the photo voltaic system, onboard AI can present a steadiness of storing knowledge and transmitting it again to Earth in the best quantities. As missions transfer farther from Earth and AI functionality is positioned onboard spacecraft, the numerous radiation-imbued environments create vital challenges for industrial off-the-shelf {hardware}, necessitating extremely customized implementations, like GSI’s radiation-tolerant APU.

About GSI Technology

Founded in 1995, GSI Technology, Inc. is a number one supplier of semiconductor reminiscence options. GSI’s assets are at the moment targeted on bringing new merchandise to market that leverage current core strengths, together with radiation-hardened reminiscence merchandise for excessive environments, and Gemini, the APU designed to ship efficiency benefits for numerous synthetic intelligence functions. GSI Technology is headquartered in Sunnyvale, California and has gross sales places of work within the Americas, Europe, and Asia.  For extra data, please go to

Source: GSI Technology


Please enter your comment!
Please enter your name here