Heterogeneous Rad-Hard Computing System with Dual Processor Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Radiation-hardened (rad-hard) devices, particularly those by design (RHBD), face limitations in radiation resilience due to non-hardened underlying components, leading to potential errors and higher power requirements, while rad-hard-by-process (RHBP) devices are constrained by older manufacturing processes resulting in lower performance and efficiency.
Innovation Solution
A rad-hard computing system comprising a rad-hard processor and a second processor, where the rad-hard processor performs critical tasks and checks the second processor's operating state and results for acceptability, allowing it to mitigate radiation-induced errors by rerouting or correcting compute-intensive tasks, thereby maintaining system reliability and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If RHBD devices employ redundant components to resist radiation, then radiation resilience is improved, but power requirements increase and energy efficiency decreases
Solution Approach 1:
The system applies different quality levels to different processors: the first processor is rad-hardened for critical tasks requiring high reliability, while the second processor uses conventional components for compute-intensive tasks where some error tolerance is acceptable. This local differentiation allows the system to achieve radiation resilience only where absolutely necessary, avoiding the power penalty of making all components redundant.
2Reliability
If RHBP devices use older semiconductor manufacturing process nodes, then radiation hardness is improved, but clock speeds decrease and compute task performance is reduced
Solution Approach 1:
The computing system is segmented into two distinct processing units: a rad-hardened first processor manufactured on older, more radiation-resistant process nodes for critical control functions, and a conventional second processor manufactured on leading-edge nodes for high-speed compute-intensive tasks. This segmentation allows each processor to be optimized for its specific function without compromise.
3Reliability
If a single rad-hardened processor performs all computing tasks, then radiation resilience is improved, but productivity and compute task performance decrease
Solution Approach 1:
The system divides computing tasks between two processors with different characteristics: the rad-hardened first processor handles critical control tasks, while the conventional high-performance second processor handles compute-intensive workloads. This segmentation enables the system to maintain radiation resilience for critical functions while achieving high productivity for non-critical computations.
Solution Approach 2:
The system uses an intermediary mechanism (task distribution and result verification logic) that allows the conventional second processor to execute compute-intensive tasks and have its results verified or corrected by the rad-hardened first processor. This intermediary layer enables the system to leverage the speed of conventional processors while maintaining the reliability guarantees of rad-hardened processing.
Data Source
AI summary
A radiation-hardened (rad-hard) computing system can include a rad-hard processor, a second processor, and a memory operably coupled to the rad-hard processor. The rad-hard processor can cause the second processor to execute compute-intensive tasks that can be performed more quickly or with fewer computing resources by the second processor than the rad-hard processor. The rad-hard processor can selectively cause the second processor to execute a compute-intensive task based upon determining that the second processor is in a normal operating state. The rad-hard processor can further evaluate computing results generate by the second processor to determine whether such computing results meet an applicable acceptability condition. The rad-hard processor can employ a computing result that meets an applicable acceptability condition in further computing operations.