Enabling Effective Error Mitigation in Memory Chips That Use On-Die Error-Correcting Codes

TL;DR

This paper investigates how on-die ECC in memory chips obscures error characteristics, proposing new testing techniques to recover and understand these errors for improved mitigation and system reliability.

Contribution

It provides a detailed analysis of on-die ECC obfuscation effects and introduces novel testing methods to reveal underlying memory error behaviors.

Findings

On-die ECC significantly obfuscates error characteristics.

New testing techniques can recover error statistics despite ECC.

Understanding error mechanisms enables better error mitigation strategies.

Abstract

Improvements in main memory storage density are primarily driven by process technology scaling, which negatively impacts reliability by exacerbating various circuit-level error mechanisms. To compensate for growing error rates, both memory manufacturers and consumers use error-mitigation mechanisms that improve manufacturing yield and allow system designers to meet reliability targets. Developing effective error mitigations requires understanding the errors' characteristics (e.g., worst-case behavior, statistical properties). Unfortunately, we observe that proprietary on-die Error-Correcting Codes (ECC) used in modern memory chips introduce new challenges to efficient error mitigation by obfuscating CPU-visible error characteristics in an unpredictable, ECC-dependent manner. This dissertation builds a detailed understanding of how on-die ECC obfuscates the statistical properties of main memory error mechanisms using a combination of real-chip experiments and statistical analyses. We experimentally study memory errors, examine how on-die ECC obfuscates their statistical characteristics, and develop new testing techniques to overcome the obfuscation. Our results show that the obfuscated error characteristics can be recovered using new memory testing techniques that exploit the interaction between on-die ECC and the statistical characteristics of memory error mechanisms to expose physical cell behavior. We conclude by discussing the critical need for transparency in DRAM reliability characteristics in order to enable DRAM consumers to better understand and adapt commodity DRAM chips to their system-specific needs. We hope and believe that the analysis, techniques, and results we present in this dissertation will enable the community to better understand and tackle current and future reliability challenges as well as adapt commodity memory to new advantageous applications.