Packet Tunneling for Multi-Node Processor Interconnects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multi-processor systems, the limited number of node identifiers (NIDs) and the need for packet routing through node controllers lead to increased latency and reduced performance due to lower bandwidth node controller links, which hinder efficient communication between processor sockets across different nodes.
Innovation Solution
Implementing packet tunneling between processor nodes without consuming additional NIDs and without going through node controller interconnects, allowing direct high-bandwidth connections between sockets and node controllers, thereby reducing latency and increasing bandwidth by altering source and destination NIDs in packets for proper addressing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If packet routing through node controllers is used to enable communication between processor sockets across different nodes, then system connectivity and communication capability are improved, but latency increases and bandwidth decreases due to lower bandwidth node controller links
Solution Approach 1:
The patent uses node identifiers as intermediaries to enable direct packet routing between processor sockets across nodes. By embedding NID information in packets and using NID translation at socket level, the system bypasses node controllers for inter-node communication while maintaining proper addressing and routing capabilities.
Solution Approach 2:
The patent segments the communication path by separating the routing function from the node controller. Individual sockets perform NID translation and routing decisions locally based on destination NIDs, dividing the communication task into socket-level routing and node-level processing, thereby avoiding the bottleneck of node controller links.
2Adaptability or versatility
If packet routing through node controllers is used to enable communication between processor sockets across different nodes, then system connectivity is improved, but bandwidth decreases due to lower bandwidth node controller links
Solution Approach 1:
Node identifiers serve as intermediaries that enable direct high-bandwidth socket-to-socket communication. The NID embedded in packets allows sockets to directly route packets across nodes using high-speed interconnects, bypassing the lower-bandwidth node controller links while maintaining proper address translation and routing.
Solution Approach 2:
The communication path is segmented to bypass node controllers for data transmission. By enabling sockets to perform local NID translation and direct routing, the patent separates control functions (remaining at node controllers) from data transmission paths (direct socket-to-socket), thereby achieving high bandwidth through direct links.
3Adaptability or versatility
If processor sockets with higher NID capabilities are used to support more nodes, then system scalability and NID availability are improved, but system costs increase
Solution Approach 1:
The patent enables sockets to perform self-service NID translation based on destination NIDs in incoming packets. Each socket maintains translation information and performs local NID mapping, eliminating the need for expensive high-NID-capability processors and reducing system costs while maintaining full scalability.
Data Source
AI summary
Multi-node, multi-socket computer systems and methods provide packet tunneling between processor nodes without going through a node controller link. On receiving a packet, the destination node identifier (NID) is examined, and if it is not same as the source socket, then the packet request address is examined. If it is determined that the packet is not for a remote connected socket, then the packet's destination NID and source socket NID are replaced along with recalculated data protection information. The modified packet is then sent to the destination socket over another processor interconnect path.