Explore the complexities of adapting software for persistent memory in this 43-minute conference talk from Strange Loop. Delve into four key ideas that have emerged from recent experiences in enabling software for persistent memory. Learn why persistent memory is neither exactly like traditional memory nor storage, and understand its role alongside other memory and storage technologies. Discover various enablement models for persistent memory and how it resurfaces classic programming challenges. Gain insights into position independence, memory allocation, garbage collection, transactions, locking, and error handling specific to persistent memory environments. Benefit from the expertise of Rob Dickinson, a systems architect and software engineer from Intel's NVML development team, as he shares valuable guidelines for developing effective persistent memory enablement strategies.
Overview
Syllabus
PROMISE AND PITFALLS OF PERSISTENT MEMORY
I'M NOT ONE OF THOSE INTEL ENGINEERS INA BUNNY SUIT
PROGRAMMING FOR PERSISTENT MEMORY
ADAPTING SOFTWARE FOR PERSISTENT MEMORY
POSITION INDEPENDENCE Virtual memory ranges change all the time Persistent memory ranges do not
MEMORY ALLOCATION & GC Regular memory allocators aren't appropriate A leak in persistent memory remains leaked
TRANSACTIONS & LOCKING Preventing torn updates using transactions Resetting abandoned locks
ERROR DETECTION & HANDLING Coping with media failures or corruption
Taught by
Strange Loop Conference
