Out-of-order executive bus dependency request links
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In out-of-order executive bus systems, traditional methods lead to delayed transactions and data hazards due to unordered request processing, resulting in decreased system performance, as seen in the '549 patent's FIFO-based solution which pauses all reading operations even when no hazards are present.
Innovation Solution
An operating method that forms dependency request links based on priority using dependency constraints, allowing requests to be processed in order and avoiding unnecessary pauses, thereby enhancing system performance and reducing data hazards.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If FIFO-based pausing method is used to avoid data hazards, then data hazard prevention is improved, but system performance deteriorates due to unnecessary pauses of all reading operations
Solution Approach 1:
The patent segments the request processing into independent permissive dependencies and strict dependencies. Permissive dependencies allow out-of-order execution without pausing, while strict dependencies enforce ordering only when necessary. This segmentation resolves the contradiction by eliminating unnecessary pauses (improving performance) while maintaining required ordering constraints (preserving reliability).
Solution Approach 2:
The patent introduces dynamic dependency tracking where the system adaptively determines whether to pause based on actual dependency relationships between requests. The pause decision is not static but dynamically adjusted based on the specific request sequence and dependency type, allowing the system to maintain reliability only when necessary and improve performance when pausing is not required.
2Reliability
If all reading operations are paused to prevent data hazards, then data hazard prevention is improved, but transaction throughput deteriorates
Solution Approach 1:
The patent divides reading operations into two categories: those that must pause (when strict dependency exists with pending write) and those that can proceed (when no strict dependency or permissive dependency only). This segmentation allows throughput to be maintained for non-critical operations while ensuring data hazard prevention for critical operations.
Solution Approach 2:
The patent changes the parameter of request dependency from a single uniform constraint to multiple types (permissive vs. strict). This parameter change enables fine-grained control over when pausing is necessary, allowing the system to maintain high throughput by only pausing when strict dependency parameters indicate a real data hazard risk.
3Stability of the object's composition
If in-order executive bus architecture is used, then request processing order is maintained, but system scalability deteriorates for large-scale systems
Solution Approach 1:
The patent makes the request execution order dynamic rather than statically fixed. Requests can execute out-of-order when permissive dependencies allow, but automatically maintain order when strict dependencies require it. This dynamic adaptability enables the system to scale efficiently while maintaining correctness, resolving the contradiction between order stability and system scalability.
Solution Approach 2:
The patent introduces dependency type parameters (permissive/strict) that control execution ordering behavior. This parameterization allows the system to adapt its ordering behavior based on the specific request characteristics, enabling scalability for large-scale systems while maintaining required ordering constraints for specific operations.
4Productivity
If out-of-order request processing is implemented, then system performance is improved, but data hazard risks increase
Solution Approach 1:
The patent segments dependencies into permissive (allowing out-of-order) and strict (requiring in-order) types. This segmentation enables the system to exploit out-of-order execution for performance improvement when permissive dependencies exist, while automatically preventing data hazards through strict dependency enforcement, thus resolving the contradiction between performance and reliability.
Solution Approach 2:
The patent introduces dependency analysis as an intermediary mechanism between out-of-order request processing and data hazard prevention. This intermediary evaluates each request's dependency type and controls execution accordingly, allowing performance benefits of out-of-order processing while filtering out data hazard risks through intelligent mediation.
Data Source
AI summary
An operating method applied to an out-of-order executive bus system includes: according to dependency constraints, linking requests using the bus system to form dependency request links having an order; and processing the order of the requests according to the dependency request links. In addition, a bus system is provided. The bus system includes a request queue and a dependency request link generator. The request queue receives and stores a newly received request including at least a link tag. The dependency request link generator generates N dependency request links according to dependency constraints of N link tags of the newly received request, where N is any positive integer. Each link tag of the newly received request is implemented to indicate a link relation with respect to an order of the newly received request and a plurality of unserved requests preceding the newly received request.