Order-Independent Memory Write Signatures for Error Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data processing devices face challenges in detecting errors caused by ionizing radiation and hardware issues, as error detection codes may fail to detect errors outside memory or in execution units, leading to undetected changes and requiring costly comparison methods between separate machines.
Innovation Solution
A data processing device with signature generation circuitry that updates a compact signature independently of write order, allowing for error detection by comparing signatures between machines, reducing overhead and enabling efficient error detection across multiple machines.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If error detection codes are stored alongside message data in memory, then errors in memory cells can be detected, but errors outside memory (in execution units) remain undetected
Solution Approach 1:
The patent divides error detection into two separate segments: error detection codes for memory data and a separate signature for execution unit results. This segmentation allows each mechanism to specialize in detecting errors in its specific domain, thereby improving overall reliability and coverage without requiring a single complex system to handle all error types.
Solution Approach 2:
The patent introduces a signature as an intermediary element that captures the state of execution unit results. This signature acts as a mediator between the execution unit and the error detection mechanism, enabling errors in the execution unit to be detected without requiring direct monitoring of all execution operations.
2Reliability
If two copies of the application are run on separate machines and results are compared, then errors can be detected, but processing time is significantly increased
Solution Approach 1:
The patent extracts only the essential error-detecting information (the signature) from the full application results. By taking out only the critical signature data for comparison rather than comparing entire application outputs, the system maintains error detection capability while dramatically reducing the time and resources required for verification.
Solution Approach 2:
The patent applies partial action by performing only the necessary comparison of signatures rather than complete result verification. This partial comparison is sufficient to detect errors while avoiding the excessive processing time that would result from comparing all application outputs in full detail.
3Reliability
If error detection is performed by comparing results from separate machines, then errors can be detected, but the overhead and time required for comparison increases significantly
Solution Approach 1:
The patent extracts only the essential error-detecting information (the signature) from the full application results. By taking out only the critical signature data for comparison rather than comparing entire application outputs, the system maintains error detection capability while dramatically reducing the time and resources required for verification.
Solution Approach 2:
The patent changes the parameter being compared from full application results to condensed signatures. This parameter transformation reduces the data volume requiring comparison from potentially megabytes to compact signature values, thereby maintaining error detection reliability while minimizing the time loss associated with verification.
Data Source
AI summary
Signature generation circuitry is configured to update a signature in response to each of a plurality of writes to memory. The signature is updated by performing bitwise operations between current bit values of the signature and at least some of the bits written to memory in response a write. The bitwise operation are order-independent such that the resulting signature is the same irrespective of the order in which the writes are used to update the signature. The signatures are formed in an order-independent manner such that, if no errors have occurred in generating the data to be written to be memory, the signatures will match. In this way, a compact signature is developed that is suitable export from the data processing device for checking against a corresponding data processing device of a machine running a duplicate application.