Programmable Cache Error Action Control Register
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Soft errors in caches due to ionizing radiation cause reliability issues in safety-critical systems, as existing technologies lack flexible mechanisms to handle and mitigate these errors effectively.
Innovation Solution
A user-programmable cache error action control register allows dynamic selection of error actions, enabling transparent operation or error reporting, thereby addressing soft errors in caches through programmable error handling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If error detection and handling mechanisms are implemented in cache, then reliability is improved, but device complexity increases
Solution Approach 1:
The patent implements a programmable error action control register that allows the error handling behavior of the cache to be dynamically changed via software. The register can be programmed to select different error actions (transparent operation, non-transparent operation with reporting, or a combination), enabling the system to adapt its error handling strategy based on operational requirements without hardware redesign.
Solution Approach 2:
The patent changes the operational parameters of the cache by introducing a control register that modifies the error response behavior. By programming different values into the control register, the cache can switch between different error handling modes, effectively changing its operational characteristics to balance reliability and transparency based on system needs.
2Productivity
If transparent error operation is implemented, then productivity is improved, but reliability monitoring capability deteriorates
Solution Approach 1:
The control register enables dynamic switching between transparent operation mode (for maximum productivity) and non-transparent reporting mode (for maximum monitoring capability). The system can program the register to select transparent operation during normal high-performance requirements, and switch to reporting mode when reliability monitoring becomes critical.
Solution Approach 2:
The system can periodically check the error action control register and switch between transparent and reporting modes based on operational phases. During normal operation, transparent mode maintains productivity, while periodic switches to reporting mode enable reliability monitoring and evaluation.
3Reliability
If non-transparent error reporting is implemented, then reliability monitoring capability is improved, but productivity deteriorates
Solution Approach 1:
The programmable control register allows the system to dynamically select non-transparent reporting mode only when reliability monitoring is required, rather than operating in this mode continuously. This selective activation minimizes the impact on productivity while maintaining the ability to monitor errors when needed.
Data Source
AI summary
A data processing system and method of operation has a processor coupled to a cache. Cache control circuitry is coupled to the cache and performs error detection. A user programmable error action control register stores a control value for selecting a type of error action to be taken when a cache error is detected. A first value of the control value permits handling of a cache error that is transparent to the processor, and a second value permits handling of the cache error by taking an exception that is visible to the processor. Various alternate actions to a detected error, including error correction or cache line invalidation, may be taken in response to other values of the control value.