Jung, Juyoung
(2017)
IMPROVING THE PERFORMANCE OF HYBRID MAIN MEMORY THROUGH SYSTEM AWARE MANAGEMENT OF HETEROGENEOUS RESOURCES.
Doctoral Dissertation, University of Pittsburgh.
(Unpublished)
This is the latest version of this item.
Abstract
Modern computer systems feature memory hierarchies which typically include DRAM as the
main memory and HDD as the secondary storage. DRAM and HDD have been extensively used for the past several decades because of their high performance and low cost per bit at their level of hierarchy. Unfortunately, DRAM is facing serious scaling and power consumption problems, while HDD has suffered from stagnant performance improvement and poor energy efficiency. After all, computer system architects have an implicit consensus that there is no hope to improve future system’s performance and power consumption unless something fundamentally changes.
To address the looming problems with DRAM and HDD, emerging Non-Volatile RAMs (NVRAMs) such as Phase Change Memory (PCM) or Spin-Transfer-Toque Magnetoresistive RAM (STT-MRAM) have been actively explored as new media of future memory hierarchy. However, since these NVRAMs have quite different characteristics from DRAM and HDD, integrating NVRAMs into conventional memory hierarchy requires significant architectural re-considerations and changes, imposing additional and complicated design trade-offs on the memory hierarchy design. This work assumes a future system in which both main memory and secondary storage include NVRAMs and are placed on the same memory bus.
In this system organization, this dissertation work has addressed a problem facing the efficient exploitation of NVRAMs and DRAM integrated into a future platform’s memory hierarchy. Especially, this dissertation has investigated the system performance and lifetime improvement endowed by a novel system architecture called Memorage which co-manages all available physical NVRAM resources for main memory and storage at a system-level. Also, the work has studied the impact of a model-guided, hardware-driven page swap in a hybrid main memory on the application performance. Together, the two ideas enable a future system to ameliorate high system performance degradation under heavy memory pressure and to avoid an inefficient use of DRAM capacity due to injudicious page swap decisions.
In summary, this research has not only demonstrated how emerging NVRAMs can be effectively employed and integrated in order to enhance the performance and endurance of a future system, but also helped system architects understand important design trade-offs for emerging NVRAMs based memory and storage systems.
Share
| Citation/Export: |
|
| Social Networking: |
|
Details
| Item Type: |
University of Pittsburgh ETD
|
| Status: |
Unpublished |
| Creators/Authors: |
|
| ETD Committee: |
|
| Date: |
23 January 2017 |
| Date Type: |
Publication |
| Defense Date: |
7 December 2016 |
| Approval Date: |
23 January 2017 |
| Submission Date: |
8 December 2016 |
| Access Restriction: |
No restriction; Release the ETD for access worldwide immediately. |
| Number of Pages: |
113 |
| Institution: |
University of Pittsburgh |
| Schools and Programs: |
Dietrich School of Arts and Sciences > Computer Science |
| Degree: |
PhD - Doctor of Philosophy |
| Thesis Type: |
Doctoral Dissertation |
| Refereed: |
Yes |
| Uncontrolled Keywords: |
Non-volatile RAM, NVRAM, memory system, hybrid memory, storage system, SSD, page swap, page migration |
| Date Deposited: |
23 Jan 2017 20:27 |
| Last Modified: |
24 Jan 2017 06:15 |
| URI: |
http://d-scholarship.pitt.edu/id/eprint/30590 |
Available Versions of this Item
-
IMPROVING THE PERFORMANCE OF HYBRID MAIN MEMORY THROUGH SYSTEM AWARE MANAGEMENT OF HETEROGENEOUS RESOURCES. (deposited 23 Jan 2017 20:27)
[Currently Displayed]
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
 |
View Item |
![[feed]](/images/feed-icon-32x32.png){kind=icon}