Flash for Supercomputing and New Non-Volatile Memory Technologies Presented at Inaugural UC San Diego Workshop
San Diego, May 17, 2010 -- From iPods to supercomputers, non-volatile memory is making its mark on modern computing. More than 130 academic and industrial researchers from around the world recently assembled at the University of California, San Diego to present the latest research on flash and other non-volatile memories that are rapidly changing the memory landscape for computers and other electronics of all sizes.
The inaugural UCSD Non-Volatile Memories Workshop (NVMW) took place on April 11-13 2010. The two-day event was hosted by the UC San Diego Center for Magnetic Recording Research (CMRR) – which is a collaboration between UC San Diego, companies in the data storage industry and government agencies; professor Steven Swanson’s Non-Volatile Systems Laboratory (NVSL) within the computer science department at the Jacobs School of Engineering; and the UCSD Division of the California Institute for Telecommunications and Information Technology (Calit2).
Non-volatile memory, which includes the flash memory in thumb drives and iPods, is gaining in popularity in part because it is rugged (unlike a hard drive, there is no spinning disk or other moving mechanical parts that don’t do well when dropped), relatively low power, and boasts fast random access speed. While flash technologies have progressed to the point that they make sense for some commercial products, it still costs more per storage unit than the latest disk drives.
“In some sense, our workshop was devoted to exploring the possibilities for including this type of memory into a larger range of applications, including high capacity storage systems,” said Prof. Paul Siegel, Director of CMRR, Co-Chair of the non-volatile memories workshop organizing committee, and a professor in the Department of Electrical and Computer Engineering (ECE) at the Jacobs School of Engineering.
In fact, the San Diego Supercomputer Center (SDSC) at UC San Diego announced in November 2009 a five-year, $20 million grant from the National Science Foundation (NSF) to build and operate Gordon, a powerful supercomputer that employs a vast amount of flash memory.
“Flash memory is at the stage that magnetic recording was at in the late 70s or early 80s. It seems the time is ripe for really exploring many different aspects of the technology and applying new techniques to improve it and really start pushing flash in all dimensions, including capacity, transfer rates and reliability,” said Siegel.
The two-day workshop included 30 invited speakers who gave presentations on a range of topics including device technologies, channel modeling and coding, system architectures, and non-volatile memory applications. The workshop also included a technical overview of solid-state storage presented by Khaled Amer of the Storage Networking Industry Association (SNIA). The speakers included researchers from leading academic, industrial, and government research labs at Google, IBM, Microsoft, Hewlett Packard, Carnegie Mellon University, Caltech, Los Alamos National Laboratory, and UC San Diego, as well as representatives from leading companies working on non-volatile memories such as Qualcomm, Numonyx, Fusion-io, Grandis, and SanDisk.
“Non-volatile, solid-state memories represent a huge shift in the way that computers handle persistent data. Ultimately, they will have a huge impact, but fully exploiting them requires changes in many system components – hardware, software, information-management – throughout the system. We wanted to get top people from all those areas together to figure how to best use non-volatile memories. That’s what made this meeting unique,” said Steven Swanson, an Assistant Professor in UCSD’s Computer Science and Engineering Department (CSE), head of the Non-Volatile Systems Laboratory (NVSL) within the computer science department at the Jacobs School of Engineering, and Co-Chair of the NVMW organizing committee.
In bringing together scientists and engineers working on different aspects of advanced non-volatile storage devices and systems, the workshop facilitated the exchange of ideas, insights, and knowledge within a broad community of practitioners and researchers, fostering the establishment of new collaborations that can propel future progress in the design and application of non-volatile memories.
“We wanted to encourage the cross-disciplinary, outside-the-box kind of thinking that will be required to realize the full potential of these new storage technologies,” said Siegel.
The final session emphasized the richness of applications that NVM technology will impact, from mobile communications to enterprise storage to large-scale biological studies; and it included perspectives of representatives from Qualcomm, STEC, and Google.
“It was very exciting to hear about such a wide range of applications. It shows how game changing these technologies will be,” said Swanson.
The technical talks were complemented by a dinner panel addressing grand challenges in non-volatile memories research. Panel participants included researchers from Microsoft Research, IBM, SanDisk, SDSC, and the Global Information Industry Center at UCSD.
“The success of the workshop reaffirms the importance and timeliness of CMRR’s new research thrust in solid-state storage,” said Siegel. “The combined vision, expertise, and collaborative spirit of researchers at CMRR, NVSL, SDSC, and Calit2 have positioned UCSD to become the academic leader in this rapidly growing field.”
The technical program and the slides for many of the presentations are available on the NVMW Web site, and news about future NVMW events will be available on the CMRR and NVSL sites (see Related Links below).
Daniel Kane, Jacobs School of Engineering, 858-534-3262, email@example.com