| NSF Org: |
CNS Division Of Computer and Network Systems |
| Recipient: |
|
| Initial Amendment Date: | August 19, 2015 |
| Latest Amendment Date: | August 19, 2015 |
| Award Number: | 1526798 |
| Award Instrument: | Standard Grant |
| Program Manager: |
Marilyn McClure
mmcclure@nsf.gov (703)292-5197 CNS Division Of Computer and Network Systems CSE Directorate for Computer and Information Science and Engineering |
| Start Date: | October 1, 2015 |
| End Date: | September 30, 2019 (Estimated) |
| Total Intended Award Amount: | $450,000.00 |
| Total Awarded Amount to Date: | $450,000.00 |
| Funds Obligated to Date: |
|
| History of Investigator: |
|
| Recipient Sponsored Research Office: |
926 DALNEY ST NW ATLANTA GA US 30318-6395 (404)894-4819 |
| Sponsor Congressional District: |
|
| Primary Place of Performance: |
Atlanta GA US 30332-0420 |
| Primary Place of
Performance Congressional District: |
|
| Unique Entity Identifier (UEI): |
|
| Parent UEI: |
|
| NSF Program(s): | CSR-Computer Systems Research |
| Primary Program Source: |
|
| Program Reference Code(s): |
|
| Program Element Code(s): |
|
| Award Agency Code: | 4900 |
| Fund Agency Code: | 4900 |
| Assistance Listing Number(s): | 47.070 |
ABSTRACT
The past decade has seen a paradigm shift in computing platforms. Emerging handheld devices such as Smartphones and Tablets have become one of the most common devices for computing in everyday use. Energy consumption is one of the prime considerations that influence the development of mobile hand-held devices, as it determines the duration for which the device remains usable on battery power. The usage patterns for devices such as smartphones are quite different from traditional computing devices such as laptops and desktops, and users have a different set of expectation from these handheld mobile devices. For example, mobile devices are expected to be always-on for at-least a day and to start-up and respond quickly. On the other hand, laptop/desktop devices are expected to perform well under heavy use, and the expectations of fast start-up and response are not as stringent. Furthermore, in terms of application usage, mobile systems tends to have burst usage with low-levels of multitasking, while desktop usage involves more continuous long-running programs with high-levels of multitasking.
The project seeks to optimize the memory system in mobile devices to enable fast-response time while maintaining low power by levering these usage differences. It consists of a three-pronged approach: First, reducing active memory footprint size by analyzing the usage patterns of mobile users by developing a usage logging program. Second, developing intelligent error correction schemes that can reduce the refresh energy consumed by the memory system without compromising data integrity. Third, reducing the memory power by utilizing emerging Non Volatile Memory (NVM) technologies and developing data partitioning techniques that can keep critical data in NVM.
Memory power consumption continues to be one of the main limiter of the battery life of mobile platforms. In additional to the general use as smartphone and tablets, efficient mobile memory system has use in several other application domains such as surveillance, automotive, environment, military, and biomedical. The techniques developed in our proposal will be useful for these domains as well. The infrastructure on usage logging and the usage characterization will foster other studies in the area of mobile memory systems.
PUBLICATIONS PRODUCED AS A RESULT OF THIS RESEARCH
Note:
When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external
site maintained by the publisher. Some full text articles may not yet be available without a
charge during the embargo (administrative interval).
Some links on this page may take you to non-federal websites. Their policies may differ from
this site.
PROJECT OUTCOMES REPORT
Disclaimer
This Project Outcomes Report for the General Public is displayed verbatim as submitted by the Principal Investigator (PI) for this award. Any opinions, findings, and conclusions or recommendations expressed in this Report are those of the PI and do not necessarily reflect the views of the National Science Foundation; NSF has not approved or endorsed its content.
Emerging handheld devices such as Smartphones and Tablets have become one of the most common devices for computing in everyday use and the energy consumption is one of the prime considerations in the development of the system. This project addressed the energy consumption problem by reducing the energy consumption of memory systems.
First, we analyzed the usage patterns of smartphones by developing a usage log monitoring tools for mobile platforms. We also developed a simulator to emulate the activities of a smartphone. Second, as machine learning applications are becoming more prevalent in everyday usage, we have also studied image recognition and classification applications. To reduce the memory consumption problem in mobile platforms, we have proposed distributed machine learning techniques that can utilize multiple mobile platforms. By adopting model parallelism, we demonstrated that this not only increases the compute scalability but also reduces the memory footprint size of each individual device, thereby improving performance and power significantly.
Third, we investigated power-efficient means of making memory systems by reducing the metadata overheads of secure memory solutions (such as Intel SGX) using reliability-security co-design and domain-specific compression. Finally, we also developed a setup to characterize commodity DRAM at cryogenic temperatures (70 kelvin) with the aim of enabling large-scale memory systems for cryogenic computing systems and for reducing refresh overheads.
The intellectual merits that are produced from this programs are (1) understanding the memory consumption behavior associated with mobile platform usage patterns (2) utilizing model parallelism to reduce memory consumption for modern machine learning workloads, (3) techniques to reduce the power consumption of secure memories by reducing metadata overheads, and (4) characterizing commodity DRAM at cryogenic environments.
The engineering contributions of this program translated the preceding intellectual contributions into open source software artifacts to benefit the larger research and development community and enable further developments. These include (1) software infrastructure to monitor memory usages for android platforms (2) distributed inference algorithms for low computing devices. These software infrastructures are in the public domain.
Last Modified: 03/13/2020
Modified by: Hyesoon Kim
Please report errors in award information by writing to: awardsearch@nsf.gov.
