RAM Power-On Training With Error-Threshold Parameter Tuning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage devices face challenges in setting optimal operation parameters for random access memory, leading to potential malfunctions and data inconsistencies due to process variations and improper initialization.
Innovation Solution
A storage device that includes a non-volatile memory to store a pattern data chunk, which undergoes a training operation of writing and reading back M times to the random access memory, allowing the controller to update operation parameters when maximum error bits exceed a threshold, optimizing the random access memory's performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Power
If the storage device uses a separate random access memory to improve controller performance, then the controller performance is improved, but the random access memory may malfunction or produce data inconsistencies due to process variations and improper initialization
Solution Approach 1:
The patent applies preliminary action by performing a training operation before normal operation. The controller writes pattern data to the random access memory and reads it back multiple times (M times) to detect error bits. Based on the detected errors, the controller pre-adjusts operation parameters such as write leveling and read data eye settings. This preliminary adjustment ensures the random access memory is properly initialized and optimized before actual data processing begins, preventing malfunctions and data inconsistencies while maintaining performance improvements.
2Reliability
If the storage device performs a training operation M times to detect error bits and update operation parameters, then the random access memory operation reliability is improved, but the power-on initialization time is increased
Solution Approach 1:
The patent applies partial or excessive action by performing the training operation a fixed number of times (M times) rather than continuously or indefinitely. This provides a balance between thoroughness and time efficiency. The value M is chosen to be sufficient to detect typical process variations and set appropriate operation parameters, but not so large as to cause excessive delay. This partial repetition ensures reliable parameter detection while controlling the time overhead during power-on initialization.
Data Source
AI summary
A storage device may execute, upon power-on, a training operation on a pattern data chunk a number of times, and update one or more operation parameters of the random access memory when the number of maximum error bits generated during the training operation is equal to or greater than a set threshold number of bits.


