Hybrid Memory Controller for DRAM and Non-Volatile Memory Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current memory systems face challenges in balancing latency, performance, capacity, and energy usage, as they typically sacrifice the advantages of one type of memory when using another, with volatile memory offering fast access but losing data upon power removal and non-volatile memory providing long data retention but with higher latencies and error rates.
Innovation Solution
A transaction-based hybrid memory system that combines volatile DRAM with non-volatile PM, NVM, or NAND devices, using a hybrid memory controller to manage access and data transfer between DRAM and non-volatile memories, optimizing performance by interleaving synchronized and nondeterministic traffic over a shared memory bus.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Duration of action of stationary object
If non-volatile memory is used to provide large memory capacity and data retention, then data retention is improved, but read/write latency increases
Solution Approach 1:
The memory system is segmented into multiple channels, with some channels dedicated to volatile DRAM and others to non-volatile memory. This segmentation allows the system to simultaneously access fast DRAM for frequently used data while maintaining large capacity in non-volatile memory, thereby reducing overall latency while preserving data retention capabilities.
Solution Approach 2:
A hybrid memory controller acts as an intermediary between the host and the combined DRAM/NVM system. It manages data transfers between volatile and non-volatile memory, implements caching strategies, and handles memory management operations, thereby optimizing the trade-off between speed and retention without requiring host intervention.
2Loss of time
If volatile DRAM is used to provide fast access speed, then read/write latency is reduced, but data retention deteriorates
Solution Approach 1:
Data that is likely to be needed is pre-loaded into the volatile DRAM portion of the hybrid memory system before it is actually accessed. The memory controller monitors access patterns and proactively transfers data from non-volatile to volatile memory, ensuring that frequently accessed data is already in fast memory when needed, thus reducing latency while maintaining retention through selective caching.
3Productivity
If dedicated DRAM devices are used to achieve high performance, then productivity is improved, but device complexity increases
Solution Approach 1:
The hybrid memory controller is designed to universally manage both volatile DRAM and non-volatile memory devices through a unified interface. It handles multiple memory types, implements various caching strategies, and provides a single point of control for the host system, thereby simplifying the overall system architecture while maintaining high performance through optimized memory management.
Data Source
AI summary
A transaction-based hybrid memory device includes a host memory controller to control operation of the device. A hybrid memory controller is coupled to the host memory controller over a memory bus. The hybrid memory controller includes non-volatile memory control logic to control operation of non-volatile memory devices and cache control logic to accelerate cache operations, a direct memory access (DMA) engine to control volatile cache memory and to transfer data between non-volatile memory, and cache memory to off load host cache managements and transactions. A host interface couples the host memory controller to the memory bus.