Demystifying Sensing Techniques for Non-Volatile Memories

Speaker:  Swaroop Ghosh – State College, PA, United States
Topic(s):  Architecture, Embedded Systems and Electronics, Robotics

Abstract

Research, development and commercialization of emerging Non-Volatile Memories (NVMs) are being aggressively pursued by the design community to supplement and/or substitute the conventional volatile and non-volatile memory technologies that are facing scalability challenges. The emerging NVMs such as, MRAM, STTRAM, RRAM and PCM are already available in market as discreet chips. According to Yole, emerging NVM market will continue to grow and find several niche applications in healthcare, banking and day-to-day computing. One of the key aspects of the functioning of these NVMs is sensing. The well-known Flash sensing techniques which are latency-intensive are not suitable for emerging NVMs. The sensing techniques for conventional charge-based memories such as, SRAM and DRAM cannot be extended to emerging NVMs since majority of these memories are resistive in nature. The emerging NVMs face stiff sensing challenges due to process, voltage and temperature variations, and, low operating voltage requirements. The desire to achieve high memory density places further restrictions on sense margin. Many conflicting trade-offs exist to maximize the sense margin at the cost of area, power and delay. A focused treatment on the sensing challenges associated with emerging NVMs is required to accelerate their research and development efforts.

This talk will cover the sensing of emerging NVMs spanning spintronic, resistive, phase change and ferroelectric memory technologies. Challenges such as, shrinking margin between resistive states due to intrinsic and extrinsic sources of variations, noise and need for large memory capacity at tighter footprint, lower operating voltages, and, faster access latency, will be discussed in detail in the context of sense amplifier design. Various state-of-the-art resilience enhancement techniques will also be presented. The talk will also investigate the impact of application e.g., in-memory computation that poses new challenges and offers new opportunities for reliable sensing.

About this Lecture

Number of Slides:  50
Duration:  60 minutes
Languages Available:  English
Last Updated: 

Request this Lecture

To request this particular lecture, please complete this online form.

Request a Tour

To request a tour with this speaker, please complete this online form.

All requests will be sent to ACM headquarters for review.