Research on the Architecture and Implementation of In-Memory Computing
Rongkai Liu, Xiancheng Lin, Xiang Gao
Abstract
The rapid growth of artificial intelligence and the bottleneck of computing force development have promoted the emergence of new computing architectures. In-memory computing, as a structure that breaks the traditional separation of memory and computing, is considered to be a method to break the bottleneck of computing force. This paper presents two typical parts of architecture design for in-memory computing, IM-A and IM-P. And it expounds the implementations of in-memory computing in various storage media. The challenges of in-memory computing in manufacture, design, and application are summarized. The development trend of in-memory computing is mainly to realize artificial intelligence algorithms.
Topics & Concepts
BottleneckComputer scienceIn-Memory ProcessingComputing with MemoryImplementationParallel computingComputer architectureReservoir computingMemory managementArchitectureDistributed computingUnconventional computingMemory architectureUniform memory accessEmbedded systemArtificial intelligenceArtificial neural networkOperating systemSemiconductor memorySoftware engineeringRecurrent neural networkInformation retrievalQuery by ExampleArtVisual artsWeb search querySearch engineAdvanced Memory and Neural ComputingFerroelectric and Negative Capacitance DevicesParallel Computing and Optimization Techniques