Abstract
The increased capacity of multi-/triple-level cells (mlc/tlc) in phase change memory (pcm) comes at the cost of higher write latency and energy, primarily due to consecutive short programming pulses in the pcm program-And-verify (p&v) approaches. l3ep is a regression-based low latency, low energy tlc pcm p&v solution. l3ep reaches the target tlc state in just one (at most five) pulse(s) for up to 53% (>95%) of our comprehensive Monte Carlo (mc) simulations. l3ep also accelerates the naturally slow pcm crystallization process by packing crystallization pulses separated by short idle periods, and augmenting them with a single verify step. l3ep thus avoids unnecessary p&v verification steps resulting in faster crystallization. Results indicate that l3ep reduces the mean p&v latency (energy) by 2.4-15× (1.9-12.2×) in comparison to state-of-The-Art p&v approaches. l3ep is configurable through a comprehensive framework to compute critical p&v parameters to customize l3ep for multiple technology nodes.
Original language | English |
---|---|
Title of host publication | Proceedings of the IEEE/ACM International Symposium on Nanoscale Architectures, NANOARCH 2017 |
Publisher | Institute of Electrical and Electronics Engineers Inc. |
Pages | 27-32 |
Number of pages | 6 |
ISBN (Electronic) | 9781509060368 |
DOIs | |
State | Published - 28 Sep 2017 |
Externally published | Yes |
Publication series
Name | Proceedings of the IEEE/ACM International Symposium on Nanoscale Architectures, NANOARCH 2017 |
---|
Bibliographical note
Funding Information:This research was supported by King Fahd University of Petroleum and Minerals and by NSF Award CCF-1217738. Corresponding author:
Publisher Copyright:
© 2017 IEEE.
ASJC Scopus subject areas
- Hardware and Architecture
- Computer Networks and Communications