Cost-Effective Reliable MLC PCM Architecture Using Virtual Data Based Error Correction

The existing charge-based memories like DRAM and flash are reaching their scaling limits. Phase change memory (PCM) is one of the most promising emerging memory technologies due to its good scalability and low leakage power. Multi-level cell (MLC) operation in PCM, which stores two or more bits in a...

Full description

Bibliographic Details
Main Authors: Taehyun Kwon, Muhammad Imran, Joon-Sung Yang
Format: Article
Language:English
Published: IEEE 2020-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/8999530/
id doaj-4cf6415432d24b66b9efa9bf83b65326
record_format Article
spelling doaj-4cf6415432d24b66b9efa9bf83b653262021-03-30T03:10:22ZengIEEEIEEE Access2169-35362020-01-018440064401810.1109/ACCESS.2020.29740138999530Cost-Effective Reliable MLC PCM Architecture Using Virtual Data Based Error CorrectionTaehyun Kwon0Muhammad Imran1Joon-Sung Yang2https://orcid.org/0000-0002-1502-5353System LSI Division, Samsung Electronics, Hwaseong, South KoreaDepartment of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, South KoreaDepartment of Systems Semiconductor Engineering, Yonsei University, Seoul, South KoreaThe existing charge-based memories like DRAM and flash are reaching their scaling limits. Phase change memory (PCM) is one of the most promising emerging memory technologies due to its good scalability and low leakage power. Multi-level cell (MLC) operation in PCM, which stores two or more bits in a single cell, is necessary to achieve a high storage density. However, a reduced resistance range in multiple storage levels of MLC PCM, introduces a lot of soft errors because of the resistance drift phenomenon. The poor reliability of MLC PCM requires strong error correction codes (ECC) which could severely degrade the storage density and performance. In this paper, we propose a cost-effective reliable MLC PCM architecture for improving MLC PCM reliability, storage density and performance. The proposed architecture exploits the data-dependent nature of resistance drift problem to reduce the ECC overhead. A simple state mapping is used to generate virtual data, which is half of the actual data size. ECC parity bits are generated based on virtual data bits instead of actual message bits, thus resulting in a reduced number of cells for parity bits. This improves the reliability and storage density of MLC PCM. The performance is also improved by minimizing the ECC overhead. Simulation results show an improvement of about 10<sup>4</sup> times in reliability and 10.9% in storage density compared to a conventional MLC PCM which uses a typical error correction scheme. Performance and energy efficiency are also improved up to 13.7% and 10%, respectively, by the proposed architecture.https://ieeexplore.ieee.org/document/8999530/Emerging memoriesphase change memory (PCM)MLC PCMresistance drifterror correction code (ECC)
collection DOAJ
language English
format Article
sources DOAJ
author Taehyun Kwon
Muhammad Imran
Joon-Sung Yang
spellingShingle Taehyun Kwon
Muhammad Imran
Joon-Sung Yang
Cost-Effective Reliable MLC PCM Architecture Using Virtual Data Based Error Correction
IEEE Access
Emerging memories
phase change memory (PCM)
MLC PCM
resistance drift
error correction code (ECC)
author_facet Taehyun Kwon
Muhammad Imran
Joon-Sung Yang
author_sort Taehyun Kwon
title Cost-Effective Reliable MLC PCM Architecture Using Virtual Data Based Error Correction
title_short Cost-Effective Reliable MLC PCM Architecture Using Virtual Data Based Error Correction
title_full Cost-Effective Reliable MLC PCM Architecture Using Virtual Data Based Error Correction
title_fullStr Cost-Effective Reliable MLC PCM Architecture Using Virtual Data Based Error Correction
title_full_unstemmed Cost-Effective Reliable MLC PCM Architecture Using Virtual Data Based Error Correction
title_sort cost-effective reliable mlc pcm architecture using virtual data based error correction
publisher IEEE
series IEEE Access
issn 2169-3536
publishDate 2020-01-01
description The existing charge-based memories like DRAM and flash are reaching their scaling limits. Phase change memory (PCM) is one of the most promising emerging memory technologies due to its good scalability and low leakage power. Multi-level cell (MLC) operation in PCM, which stores two or more bits in a single cell, is necessary to achieve a high storage density. However, a reduced resistance range in multiple storage levels of MLC PCM, introduces a lot of soft errors because of the resistance drift phenomenon. The poor reliability of MLC PCM requires strong error correction codes (ECC) which could severely degrade the storage density and performance. In this paper, we propose a cost-effective reliable MLC PCM architecture for improving MLC PCM reliability, storage density and performance. The proposed architecture exploits the data-dependent nature of resistance drift problem to reduce the ECC overhead. A simple state mapping is used to generate virtual data, which is half of the actual data size. ECC parity bits are generated based on virtual data bits instead of actual message bits, thus resulting in a reduced number of cells for parity bits. This improves the reliability and storage density of MLC PCM. The performance is also improved by minimizing the ECC overhead. Simulation results show an improvement of about 10<sup>4</sup> times in reliability and 10.9% in storage density compared to a conventional MLC PCM which uses a typical error correction scheme. Performance and energy efficiency are also improved up to 13.7% and 10%, respectively, by the proposed architecture.
topic Emerging memories
phase change memory (PCM)
MLC PCM
resistance drift
error correction code (ECC)
url https://ieeexplore.ieee.org/document/8999530/
work_keys_str_mv AT taehyunkwon costeffectivereliablemlcpcmarchitectureusingvirtualdatabasederrorcorrection
AT muhammadimran costeffectivereliablemlcpcmarchitectureusingvirtualdatabasederrorcorrection
AT joonsungyang costeffectivereliablemlcpcmarchitectureusingvirtualdatabasederrorcorrection
_version_ 1724183972073701376