Multiprocessor Node Interface Prioritizing Critical Word Transmission
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multiprocessor systems, the latency associated with receiving critical data can be problematic due to the need to maintain coherency between shared memory and onboard caches, often requiring the transmission of entire cache lines that include both critical and non-critical data.
Innovation Solution
A system and method where a first node interface receives requests for critical words of cache lines and conveys these requests to a second node interface, prioritizing the transmission of critical words before non-critical words, ensuring efficient data transfer between processing nodes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If entire cache lines are transmitted to maintain coherency, then coherency between shared memory and onboard caches is maintained, but latency for receiving critical data increases
Solution Approach 1:
The cache line is segmented into critical words and non-critical words. The node interface transmits only the critical words first before transmitting the non-critical words, allowing the receiving processor to obtain the most time-sensitive data faster while still maintaining coherency through selective transmission of cache line portions.
Solution Approach 2:
The node interface performs preliminary action by identifying and transmitting critical words before non-critical words. This prioritization ensures that the most time-sensitive data is delivered first, reducing latency for critical operations while maintaining the option to transmit the complete cache line for coherency when necessary.
2Productivity
If critical words are transmitted before non-critical words, then data access efficiency improves, but complexity of data transmission management increases
Solution Approach 1:
The node interface performs self-service by autonomously identifying critical words and non-critical words within cache lines and automatically managing the transmission priority without requiring complex external control mechanisms. The processor core itself marks words as critical, and the node interface handles the prioritized transmission automatically.
Data Source
AI summary
A system for conveying critical and non-critical words of multiple cache lines includes a first node interface of a first processing node receiving, from a first processor, a first request identifying a critical word of a first cache line and a second request identifying a critical word of a second cache line. The first node interface conveys requests corresponding to the first and second requests to a second node interface of a second processing node. The second node interface receives the corresponding requests and conveys the critical words of the first and second cache lines to the first processing node before conveying non-critical words of the first and second cache lines.