Partial Soft Bit Read for Higher-Throughput LDPC Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Low-density parity-check (LDPC) error correction mechanisms in storage devices face performance degradation due to the overhead of performing soft bit page read operations, particularly in systems with low frequency/bandwidth buses, as they require additional sensing and toggle operations that reduce overall throughput.
Innovation Solution
Implementing a partial soft bit read operation where only a portion of the soft bit data is transferred from memory to the ECC engine, determined by selecting specific ECC codewords or subcodes for transfer based on syndrome weight thresholds, thereby reducing the data transfer burden and maintaining performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a soft bit page read operation is performed to improve decoding performance, then the LDPC ECC engine receives soft bit information indicating which memory cells have high probability of being erroneous, but the additional sensing and toggle operations decrease overall throughput at the storage device
Solution Approach 1:
The patent divides the soft bit data into multiple subcodes corresponding to different ECC codewords. Instead of transferring all soft bit data uniformly, only selected subcodes are transferred based on syndrome weight thresholds. This segmentation allows the system to focus resources on the most error-prone portions of the data, improving decoding performance where needed while minimizing unnecessary data transfers that would reduce throughput.
Solution Approach 2:
The patent implements partial soft bit read by transferring only a portion of the soft bit data rather than all of it. The controller determines which subcodes to transfer based on syndrome weight calculations, transferring only those subcodes that exceed error thresholds. This partial action provides sufficient soft bit information for improved decoding while reducing the data transfer burden on the bus, thereby maintaining higher throughput compared to full soft bit page read operations.
2Loss of information
If all soft bit data is transferred from memory to the ECC engine, then complete soft bit information is available for decoding, but the data transfer burden increases and throughput decreases in systems with low frequency/bandwidth buses
Solution Approach 1:
The patent extracts only the necessary soft bit information from the complete soft bit data set. By calculating syndrome weights and comparing them against thresholds, the system identifies which subcodes contain the most critical error information. Only these extracted subcodes are transferred to the ECC engine, eliminating the need to transfer redundant soft bit data that would consume bus bandwidth without providing additional decoding benefit.
Solution Approach 2:
The patent applies different quality levels of soft bit information transfer to different portions of the data. Instead of uniform transfer, each subcode is evaluated individually based on its syndrome weight, and transfer decisions are made locally for each subcode. This allows high-priority subcodes with high error probabilities to receive full soft bit information while lower-priority subcodes receive reduced or no soft bit information, optimizing the balance between information completeness and throughput.
Data Source
AI summary
A data storage device includes a non-volatile memory and a controller coupled to the non-volatile memory. The controller includes an error correction coding (ECC) decoder. The non-volatile memory is configured to sense hard bit data and soft bit data corresponding to multiple ECC codewords from a word line of the non-volatile memory and to sense soft bit data for the multiple ECC codewords. The soft bit data includes sub codes for each of the multiple ECC codewords. The non-volatile memory is configured to send less than all of the sensed soft bit sub codes to the ECC decoder.


