In a keynote talk at the 2017 FAST conference in Santa Clara, CA, Kimberly Keeton from #HPE spoke about Memory Driven Computing. At the same time that data demands are increasing due to #bigdata, )#IoT, #bigvideo and other drivers, microprocessor improvements (especially in processing speed) are limited by the slowing up scaling (often referred to as Moore’s Law). At the same time, data access to memory is becoming a bigger factor in overall system performance. As a consequence we need to look at a transition from processor-centric computing to memory-driven computing.
The development of new non-volatile memory options with various trade-offs between performance, data retention and endurance is leading to a new memory/storage hierarchy. Magnetic RAM ( #MRAM ), Resistive RAM ( #RRAM ) and Phase Change memory ( #PCM ) products are now available on the market offering a range of latencies.
No comments:
Post a Comment