Processing in Memory Data Compression for Bandwidth Bottlenecks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional computer architectures experience increased data transfer latency, reduced data communication bandwidth, and higher energy consumption due to the distance between memory and remote processing units, especially when handling large data volumes, which hampers computational performance.
Innovation Solution
Implementing a Processing in Memory (PIM) component within the memory module to compress and decompress data locally, reducing the need for data transfer by storing compressed data in a dedicated block within the memory, thereby offloading compression and decompression tasks from the processing unit and conserving computational resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If data is transferred between memory and remote processing unit, then data communication is achieved, but data transfer latency increases and bandwidth reduces
Solution Approach 1:
The patent merges the processing unit with the memory module to create a PIM component, eliminating the need for long-distance data transfer between separate memory and processing units. This integration allows data processing to occur directly within the memory module, significantly reducing communication pathway length and improving data transfer speed.
Solution Approach 2:
The PIM component acts as an intermediary between the memory and the remote processing unit, handling data processing tasks locally before results need to be communicated back. This intermediary function reduces the volume of data that must traverse the long communication pathway between memory and remote processor.
2Use of energy by moving object
If data is transferred between memory and remote processing unit, then data communication is achieved, but data communication energy increases
Solution Approach 1:
By merging the processing unit with the memory module, the patent eliminates long-distance data transfers that consume significant energy. The PIM component processes data locally within the memory module, dramatically reducing the energy required for data communication over long pathways.
3Productivity
If compression tasks are performed by the processing unit, then data compression is achieved, but computational resources are consumed
Solution Approach 1:
The PIM component provides self-service by performing compression tasks locally within the memory module, eliminating the need for the remote processing unit to consume computational resources for compression. The data is compressed at the source without requiring additional processing power from the main CPU.
Solution Approach 2:
The patent extracts the compression function from the remote processing unit and places it within the memory module's PIM component. This extraction of the compression task from the main processor reduces computational resource consumption while maintaining compression efficiency.
Data Source
AI summary
In accordance with the described techniques for data compression and decompression for processing in memory, a page address is received by a processing in memory component that maps to a first location in memory where data of a page is maintained. The data of the page is compressed by the processing in memory component. Further, compressed data of the page is written by the processing in memory component to a compressed block device responsive to the compressed data satisfying one or more compressibility criteria. The compressed block device is a portion of the memory dedicated to storing data in a compressed form.