Lockstep Comparator Circuit Periodic Self-Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current lockstep systems face challenges in detecting faults within the comparison circuit without imposing significant software architecture limitations or violating idle-time duration constraints, which can lead to undetected faults and increased risk of system failures.
Innovation Solution
Implementing a method where at least two comparator circuits operate between compute engines to compare memory accesses and program counters, allowing for periodic non-intrusive diagnosis in three modes, enabling continuous operation and comprehensive fault detection without idle time constraints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single comparator circuit is used to compare lockstep processing systems, then device complexity is reduced, but the system cannot perform self-diagnosis without halting computation
Solution Approach 1:
The patent combines the comparison function and self-diagnosis function into a single comparator circuit by time-multiplexing. The comparator alternates between comparing lockstep outputs during computation cycles and executing self-diagnosis during idle cycles, eliminating the need for separate diagnosis circuits while maintaining both functionalities.
Solution Approach 2:
The comparator circuit operates periodically, switching between comparison mode and self-diagnosis mode based on system state. During active computation, it compares lockstep outputs; during idle periods, it executes self-diagnosis routines, enabling continuous monitoring without requiring additional hardware.
2Reliability
If the comparator circuit is halted for diagnosis, then fault detection is possible, but computation must be suspended violating idle-time constraints
Solution Approach 1:
The system performs self-diagnosis periodically during naturally occurring idle cycles rather than halting computation. The comparator executes diagnosis routines during these idle periods when no computation is occurring anyway, so system availability is maintained while still enabling fault detection.
Solution Approach 2:
The comparator circuit performs self-diagnosis autonomously during idle cycles without external intervention or computation suspension. The diagnosis is integrated into the normal operation cycle, allowing the system to self-monitor without impacting productivity.
3Reliability
If multiple comparator circuits are used for redundancy, then fault detection coverage is improved, but device complexity and cost increase
Solution Approach 1:
The patent merges multiple comparator functionalities into a single comparator circuit that time-multiplexes between comparison operations and self-diagnosis. This eliminates the need for separate dedicated diagnosis comparators while achieving equivalent or superior fault detection coverage.
Solution Approach 2:
The comparator circuit is designed to perform multiple functions: comparing lockstep outputs during computation and executing self-diagnosis during idle cycles. This multi-functionality eliminates the need for separate specialized circuits, reducing overall device complexity while maintaining comprehensive fault detection.
4Reliability
If computation is halted for self-diagnosis, then comprehensive fault detection is possible, but idle-time duration constraints are violated
Solution Approach 1:
The self-diagnosis is executed periodically during naturally occurring idle cycles in the computation flow. By aligning diagnosis execution with these pre-existing idle periods, the system achieves comprehensive fault detection without adding extra idle time or violating duration constraints.
Solution Approach 2:
The system maintains continuous useful action by performing self-diagnosis during idle cycles rather than halting computation. The comparator remains active throughout, continuously either comparing outputs or executing diagnosis, ensuring no productive time is lost while still achieving comprehensive fault detection.
Data Source
Figure 1
Figure 2A~2B
Figure 3
AI summary
Aspects disclosed herein relate to periodic non-intrusive diagnosis of lockstep systems. An exemplary method includes comparing execution of a program on a first processing system of the plurality of processing systems and execution of the program on a second processing system of the plurality of processing systems using a first comparator circuit, comparing the execution of the program on the first processing system and the execution of the program on the second processing system using a second comparator circuit, and running a diagnosis program on the second comparator circuit while the comparing using the first comparator circuit is ongoing.