Intel has reportedly won a Phase 2 order from the U.S. Department of Defense for its latest Heterogeneous Integration Prototyping (SHIP) technology. The SHIP program enables the U.S. government to leverage Intel’s state-of-the-art semiconductor packaging technology in Arizona and Oregon, and leverage the accumulation of Intel’s tens of billions of dollars in R&D and manufacturing investments each year.
SHIP is a project researched and advanced by the Office of the Under Secretary of Defense and funded by the Trusted and Assured Microelectronics program. The second phase of the program will develop prototypes of multi-chip packages and accelerate the development of interface standards, protocols and security for heterogeneous systems.
The SHIP prototype will integrate specialized government chips with Intel’s advanced commercial chip products, including field programmable gate arrays, application-specific integrated circuits and CPUs. This combination of technologies provides U.S. government industry partners with new ways to develop mission-critical systems for modern government while leveraging Intel’s U.S. manufacturing capabilities.
To ensure that the U.S. defense industrial base can continue to provide state-of-the-art electronics for national security, the Department of Defense (DoD) must partner with America’s leading semiconductor companies,” said Nicole Petta, chief director of microelectronics in the Office of the Undersecretary of Defense Research and Engineering. said.” The DoD Microelectronics Roadmap recognizes the importance of strategic partnerships with industry. The roadmap also prioritizes and recognizes that heterogeneous assembly technology is a critical investment for both the Department of Defense and our nation as the rate of process scaling decreases. SHIP directly contributes to advancing the goals outlined in the DoD roadmap, and DoD looks forward to working with Intel, the global leader in this technology. “Nicole Petta added.
So-called heterogeneous packaging allows multiple individually fabricated integrated circuit dies (chips) to be assembled on a single package, thereby increasing performance while reducing power consumption, size and weight. SHIP enables the US government to use Intel’s advanced heterogeneous packaging technologies including Embedded Multi-Chip Interconnect Bridge (EMIB), 3D Foveros and Co-EMIB (combining EMIB and Foveros).
In addition to this, Intel has entered into a partnership with Sandia National Laboratories to test the scale-up potential of neuromorphic computing.
Intel’s 2017 neuromorphic chip called Loihi, which aims to directly mimic the behavior of the human brain, has learned to smell, touch and even help children in wheelchairs. Intel is currently in the fifth generation of neuromorphic research. Earlier this year, Intel expanded Loihi into a system called Pohoiki Springs, a behemoth containing 768 Loihi chips, each with 128 cores, and about 131,000 “neurons” that simulate computing (system-wide about 100 million digital neurons in total). Pohoiki Springs is an “experimental balloon”, even a very large one, initially only available to members of the Intel Neuromorphic Research Community (INRC) via the cloud.
But Intel’s latest large-scale deployment of neuromorphic systems will be something else entirely. Through a three-year agreement with Sandia National Laboratories, Intel will provide Loihi-based systems, setting the stage for later stages of the collaboration. Learn from our large-scale study of Intel’s upcoming neuromorphic architecture and the delivery of Intel’s largest neuromorphic system. While their first system will reach about 50 million computational neurons (roughly containing 384 Loihi chips), the latter system “may be over a billion neurons in computational power”, which equates to about 7,600+ Loihi chips. “
Intel’s rapid scaling of neuromorphic computing over the past few years has signaled confidence in the new technology – given early results showing the technology is four times more energy efficient than state-of-the-art CPUs in the U.S. on Pohoiki Springs orders of magnitude, so Intel believes that confidence has been fully affirmed. Sandia’s goal is to identify areas where neuromorphic computing is best suited to help solve some of America’s most pressing problems, such as energy and national security.
“By using the high-speed, efficient, and adaptive capabilities of neuromorphic computing architectures, Sandia National Laboratories will explore the acceleration of high-demand and evolving workloads that are increasingly important to our national security,” Intel Neuromorphic Computing Experiments said the director of the chamber, Mike Davies. “We look forward to fruitful collaborations leading to the development of next-generation neuromorphic tools, algorithms and systems that can scale to the billion-neuron level and beyond.”
To keep Intel’s neuromorphic computing on track, Sandia will evaluate the scaling range of various spiking neural network workloads, from physical modeling to large-scale deep networks, that provide a good indication that the chip is suitable for particle interaction Action simulation. Sandia National Laboratories, one of three national laboratories serving the National Nuclear Security Administration (NNSA), as the manager of the national nuclear weapons stockpile, has a particular interest in particle and fluid modeling, And just announced another big supercomputer from HPE (powered by the upcoming Sapphire Rapids Xeons).
“Sandia National Laboratory has long been a leader in large-scale computing, using some of the most advanced high-performance computers in the country to improve national security. As the need for real-time, dynamic data processing becomes more pressing, We are exploring entirely new computing paradigms, such as neuromorphic architectures,” said Craig Vineyard, principal member of Sandia’s technical staff. “Our work has helped Sandia maintain its leadership in computing, and this new effort from Intel’s Neuromorphic Research Group will continue that legacy into the future.”
The Links: 2MBI100NC-120 LB104V03-TJ01 IGBTMODULE