Glossary of Terms Used on This Site

It seems the industry is mostly aligned on these terms, but not 100%. Taken literally, the terms Non-Volatile Memory and Persistent Memory would seem to be the same thing. But in contemporary usage, the terms are different. Here’s how these terms are used by this project.

Non-Volatile Memory (NVM)

This term refers to the category of solid-state storage devices, from the Flash Memory used in Solid State Disk (SSD) drives, to battery-backed up memory cards, to the emerging non-volatile technologies such as 3D XPoint, PCM, Memristor, STT-RAM, etc.

In this project we use the term NVM to refer to the entire class of non-volatile technologies. For the faster varieties, where direct, load/store access is appropriate, read on…

Persistent Memory

This term refers to memory that maintains its contents across power failure. We specifically refer to memory technologies that are fast enough so it is reasonable to stall a CPU load instruction, waiting for a load directly from persistent memory.

By this definition, the NAND Flash used in SSDs would not qualify as persistent memory when accessed directly, but battery-backed DRAM or (hopefully) the emerging non-volatile memory technologies would qualify for direct access.

Persistent Memory Development Kit

PMDK is a project with the goal of making persistent memory programming easier. It currently supports ten libraries, targeted at various use cases for persistent memory, along with language support for C, C++, Java, and Python, tools like the pmemcheck plug-in for valgrind, and an increasing body of documentation, code examples, tutorials, and blog entries. The libraries are tuned and validated to production quality and issued with a license that allows their use in both open- and closed-source products. And the project continues to grow as we learn about new use cases.

Since the programming model for persistent memory is based on memory-mapped files, the PMDK libraries will work correctly on top of any NVM technology (like an SSD with a traditional file system on it). However, it is optimized for the fine-grained, load/store persistence provided by persistent memory, so the performance will on traditional storage will be non-optimal, as operations like cache line flushes get turned into page writes to a block device on traditional storage. This means you can use storage (like your laptop’s disk) to try out the PMDK libraries and you will get correct results. But you won’t see performant results unless you use actual persistent memory.