Packed ECC for Compressed Data With End-to-End Error Coverage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional inline error correction codes (ECC) in memory devices increase memory bandwidth usage and power consumption due to separate ECC access requirements, and they lack end-to-end coverage, especially in crossbar structures and data transmission between processing and storage units.
Innovation Solution
The implementation of packed error correction codes that opportunistically combine ECC check-bits with compressed data, allowing for end-to-end error detection and correction within processors and across multiple structures, including crossbar structures, by determining the minimum number of fixed-size storage units needed to store both compressed data and check-bits, and transmitting this combination for storage and correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional inline ECC is used to protect compressed data, then error detection and correction capability is provided, but memory bandwidth usage and power consumption increase due to separate ECC access requirements
Solution Approach 1:
The patent combines ECC check-bits with compressed data into a single packed data structure that can be accessed and transmitted together. This merging eliminates the need for separate ECC access operations, thereby reducing memory bandwidth usage and power consumption while maintaining error detection and correction capability.
Solution Approach 2:
The packed data structure serves multiple functions simultaneously: it stores both compressed data and ECC check-bits, enables end-to-end error protection across memory hierarchy and transmission paths, and can be processed as a single unit by multiple components including crossbar structures and storage units.
2Reliability
If conventional inline ECC is used to protect compressed data, then error detection and correction capability is provided, but memory bandwidth usage increases due to separate ECC access requirements
Solution Approach 1:
The patent combines ECC check-bits with compressed data into a single packed data structure that can be accessed and transmitted together. This merging eliminates the need for separate ECC access operations, thereby reducing memory bandwidth usage and power consumption while maintaining error detection and correction capability.
3Reliability
If conventional inline ECC is used, then error protection is provided for memory devices, but end-to-end coverage is lacking especially in crossbar structures and data transmission between processing and storage units
Solution Approach 1:
The packed data structure serves multiple functions simultaneously: it stores both compressed data and ECC check-bits, enables end-to-end error protection across memory hierarchy and transmission paths, and can be processed as a single unit by multiple components including crossbar structures and storage units.
Solution Approach 2:
The ECC check-bits are computed and packed with the compressed data at the source point before transmission or storage. This preliminary action ensures that error protection is established in advance, allowing the same packed structure to be used throughout the entire data path including crossbar structures and storage units, thereby eliminating coverage holes.
Data Source
AI summary
A packed error correction code (ECC) technique opportunistically embeds ECC check-bits with compressed data. When compressed, the data is encoded in fewer bits and is therefore fragmented when stored or transmitted compared with the uncompressed data. The ECC check-bits may be packed with compressed data at “source” points. The check-bits are transmitted along with the compressed data and, at any “intermediate” point between the source and a “destination” the check-bits may be used to detect and correct errors in the compressed data. In contrast with conventional systems, packed ECC enables end-to-end coverage for sufficiently-compressed data within the processor and also externally. While storage circuitry typically is protected by structure-specific ECC, protection is also beneficial for data as it is transmitted between processing and/or storage units.