Intel actually started out as a DRAM memory, and once became a pioneer in storage semiconductor technology, and in 1970 successfully built the world’s first commercial 1K Bit pMOS DRAM “Intel 1103”, but with the later rise of Japanese manufacturers in 1984 announced its withdrawal from the field.
Although no longer developing DRAM technology, Intel is not completely cut off from storage technology, for example, the company later announced at IDF 2015 to cooperate with Micron to develop 3D XPoint technology, put efforts into the field of NAND Flash, and also developed Optane Aeon hard disk, but in a few years, the related business was sold off.
Intel subsidiary “Intel Federal LLC” accepted the commission of the U.S. Department of Energy to develop a new memory technology for Sandia National Laboratories. According to the report, this technology is part of the U.S. AMT Advanced Memory Technology, mainly used in the field of supercomputing, designed to accelerate simulation and computing applications.
No definitive information is available on this new DRAM technology, but several technologies have been selected and overall appear to be very powerful, up to 40 times more effective than those used in Sandia Labs’ upcoming NNSA supercomputer.
The project is funded by the National Nuclear Security Administration’s Advanced Simulation and Computing Program, in partnership with Sandia, Los Alamos, and Lawrence Livermore national laboratories, and is part of NNSA’s post-trillion sub-computing program portfolio, whose goal is to maintain technology development momentum and improve U.S. industry competitiveness in next-generation high-performance computing technologies through close collaboration with industry initiated by its PathForward program The goal is to maintain technology development momentum and to foster a stronger domestic HPC ecosystem through close collaboration with industry initiated by its PathForward program.
“This work will focus on improving the bandwidth and latency characteristics of future memory systems, which will have a direct impact on application performance across a wide range of ASC task codes,” said James H. Laros III, Sandia project leader.
“We are already anticipating the next generation of future platform challenges that must be addressed,” said Anil Rao, vice president and general manager of Intel’s Systems Architecture and Engineering Group, “and we believe the Advanced Memory Technology Program will help us support the next decade of innovation.”
“We are rethinking the way DRAM is organized and the fundamental aspects of how it can be combined with computing platforms to achieve breakthrough performance,” said Intel Fellow Josh Fryman. “We intend to fundamentally advance computer system architecture by examining the toughest questions posed by scientists at Sandia, Lawrence Livermore and Los Alamos National Laboratory. Mainstream memory is not designed for today’s computing platforms, and this multi-year effort will help us achieve orders of magnitude performance gains from the basic DRAM design itself, leading to entirely new levels of performance across all industry sectors. We expect to see these innovations incorporated into industry standards to enhance the entire ecosystem.”