Memory Banks With In-Memory Operators to Reduce Data Transfer
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing computing systems face challenges in efficiently processing compute-intensive applications like graphics algorithms and neural network processing due to the large amount of data transmission between memory devices and processors, necessitating improved memory management and operation efficiency.
Innovation Solution
Incorporation of in-memory operators in memory banks to encode, perform operations on, and decode data, along with a memory controller to manage these processes, utilizing sparsification and quantization compression to reduce data size and enhance memory efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is transmitted between memory device and processor for compute-intensive operations, then processing capability is improved, but data transmission volume increases
Solution Approach 1:
The patent merges the memory storage function with the data processing function by implementing in-memory operators that can perform operations directly within the memory device. This combines data storage and computation in a single location, eliminating the need for frequent data transmissions between memory and processor while maintaining processing capability.
Solution Approach 2:
The patent applies preliminary encoding and compression operations on data within the memory device before it is needed for processing. The in-memory operator encodes data using sparsification and quantization techniques, reducing its size and optimizing it for subsequent operations, thereby reducing the quantity of data that needs to be transmitted.
2Quantity of substance
If in-memory operator encodes data using sparsification and quantization, then data size is reduced, but encoding complexity increases
Solution Approach 1:
The memory device performs encoding operations autonomously using its own internal resources. The in-memory operator utilizes the memory device's existing computational capabilities to perform sparsification and quantization without requiring external encoding hardware or complex external processing systems, thereby reducing overall system complexity despite the advanced encoding techniques used.
3Productivity
If in-memory operator performs operations on encoded data, then operation efficiency is improved, but memory usage increases
Solution Approach 1:
The system performs encoding compression as a preliminary action before storing and operating on data. By compressing data into a more compact representation through sparsification and quantization, the memory device can store and process data more efficiently, reducing the actual memory space required while maintaining operation efficiency.
Data Source
AI summary
A memory device and an operation method performed by the memory device are disclosed. The memory device includes a plurality of memories, and one or more memory banks including an in-memory operator configured to encode data stored in at least one of the plurality of memories, perform an assigned operation based on the encoded data, and decode the encoded data on which the operation is performed, and a memory controller configured to control the one or more memory banks.