Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Towards Successful Application of Phase Change Memories: Addressing Challenges from Write Operations

Zhou, Ping (2012) Towards Successful Application of Phase Change Memories: Addressing Challenges from Write Operations. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

[img]
Preview
PDF (ETD updated on 3/18/2012 according to guidelines)
Primary Text

Download (2MB) | Preview

Abstract

The emerging Phase Change Memory (PCM) technology is drawing increasing attention due to its advantages in non-volatility, byte-addressability and scalability. It is regarded as a promising candidate for future main memory. However, PCM's write operation has some limitations that pose challenges to its application in memory. The disadvantages include long write latency, high write power and limited write endurance.

In this thesis, I present my effort towards successful application of PCM memory. My research consists of several optimizing techniques at both the circuit and architecture level. First, at the circuit level, I propose Differential Write to remove unnecessary bit changes in PCM writes. This is not only beneficial to endurance but also to the energy and latency of writes. Second, I propose two memory scheduling enhancements (AWP and RAWP) for a non-blocking bank design. My memory scheduling enhancements can exploit intra-bank parallelism provided by non-blocking bank design, and achieve significant throughput improvement. Third, I propose Bit Level Power Budgeting (BPB), a fine-grained power budgeting technique that leverages the information from Differential Write to achieve even higher memory throughput under the same power budget. Fourth, I propose techniques to improve the QoS tuning ability of high-priority applications when running on PCM memory.

In summary, the techniques I propose effectively address the challenges of PCM's write operations. In addition, I present the experimental infrastructure in this work and my visions of potential future research topics, which could be helpful to other researchers in the area.


Share

Citation/Export:
Social Networking:
Share |

Details

Item Type: University of Pittsburgh ETD
Status: Unpublished
Creators/Authors:
CreatorsEmailPitt UsernameORCID
Zhou, Pingpiz7@pitt.eduPIZ7
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairYang, Junjuy9@pitt.eduJUY9
Committee CoChairZhang, Youtaozhangyt@cs.pitt.eduYOUTAO
Committee MemberChilders, Brucechilders@cs.pitt.eduCHILDERS
Committee MemberCho, Sangyeuncho@cs.pitt.eduSANGYEUN
Committee MemberLevitan, Stevenlevitan@pitt.eduLEVITAN
Committee MemberYun, Minheemiy16@pitt.eduMIY16
Date: 4 June 2012
Date Type: Publication
Defense Date: 22 November 2011
Approval Date: 4 June 2012
Submission Date: 3 January 2012
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 143
Institution: University of Pittsburgh
Schools and Programs: Swanson School of Engineering > Computer Engineering
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: Phase Change Memory, memory scheduling, power budgeting
Date Deposited: 04 Jun 2012 18:31
Last Modified: 15 Nov 2016 13:55
URI: http://d-scholarship.pitt.edu/id/eprint/10870

Metrics

Monthly Views for the past 3 years

Plum Analytics


Actions (login required)

View Item View Item