Direct-Link Computing Processor with Local Routing Tables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large-scale distributed computing systems experience high system latency and increased complexity and cost due to the use of general-purpose processors and dedicated network switches, which hinder efficient data processing and communication between computing nodes and clusters.
Innovation Solution
A unified and directly connected computing system is implemented, where computing processors are connected through direct links both within and between computing nodes and clusters, and each processor maintains a simple local routing table for efficient data routing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If general-purpose processors and dedicated network switches are used for routing communication requests, then routing functionality is provided, but system latency increases and processing efficiency decreases
Solution Approach 1:
The patent merges the routing functionality from separate network switches into the computing processors themselves. Each computing processor is equipped with a routing table and can directly route communication requests to other computing processors within the same computing node, eliminating the need for dedicated network switches and reducing system latency while maintaining routing functionality.
Solution Approach 2:
The patent extracts the routing function from the network switch and relocates it to the computing processor. By taking out the routing capability from the dedicated network hardware and embedding it in the computing processor's software/firmware, the system eliminates unnecessary communication hops and reduces latency.
2Reliability
If general-purpose processors and dedicated network switches are deployed, then communication routing is enabled, but hardware complexity and system cost increase
Solution Approach 1:
The patent combines multiple functions (computing and routing) into a single component (the computing processor). By merging the routing function with the computing processor, the system reduces the number of separate hardware components needed, thereby simplifying the overall hardware architecture and reducing system cost while maintaining full routing capability.
Solution Approach 2:
The computing processor is designed to perform multiple functions: both computational tasks and routing operations. By making the computing processor universal and capable of handling both data processing and network routing, the system eliminates the need for dedicated network switches, reducing hardware complexity and cost.
3Reliability
If communication requests are routed through network switches using global routing tables, then system-wide routing is achieved, but communication consumption and processing overhead increase
Solution Approach 1:
The patent segments the routing function into local routing tables at each computing processor rather than using a single global routing table. Each computing processor maintains its own routing table for routing to other processors within the same computing node, which reduces the size and complexity of routing tables and decreases communication consumption for routing operations.
Solution Approach 2:
The patent implements local routing tables at each computing processor that contain routing information specific to the local computing node. This local quality approach allows routing decisions to be made with minimal information, reducing communication overhead and energy consumption compared to querying or maintaining global routing tables.
Data Source
AI summary
The present disclosure provides a computing system, a computing processor and a data processing method for the computing processor. The computing system includes: multiple computing clusters, each computing cluster includes multiple computing nodes, and each computing node includes multiple computing processors. At least some computing clusters among the computing clusters, at least some computing nodes in each computing cluster and at least some computing processors of each computing node are connected through direct links. Each computing processor of at least some computing processors of the computing node is configured with a local routing table, which is configured for the computing processor to determine, based on the local routing table, the next direct link through which a data packet performs routing from a data source to a data destination, and the computing processor forwards the data packet through the next direct link.