High-Performance Software Techniques for Persistent Memory
Speaker: Yan Solihin – Orlando, FL, United StatesTopic(s): Architecture, Embedded Systems and Electronics, Robotics
Abstract
Persistent memory is being integrated into the main memory of various computer systems, due to its low cost, scaling potential, and non-volatility. I will present the benefit of making data structures persistent in memory, including avoiding the use of files, and as alternative way for fault tolerance. I will also discuss Lazy Persistency, a software technique that allows code to run free of logging and persist barrier overheads. Upon a crash, checksum is used to detect persistency failure and re-execute code that failed to persist. Lazy Persistency achieves performance comparable to native execution and does not require hardware instruction support or any persistency models, hence it can be readily deployed. Finally, I will talk about how memory encryption and integrity verification are difficult to achieve with NVMM, and present a few methods to mitigate the high performance overheads.About this Lecture
Number of Slides: 50Duration: 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.