Hop-by-hop error detection in a server system
AssigneeAdvanced Micro Devices, Inc.
InventorsMin Xu, Sean Lie, Gene Shen
Life of this patent
Estimating the term…
Abstract
A server system performs error detection on a hop-by-hop basis at multiple compute nodes, thereby facilitating the detection of a compute node experiencing failure. The server system communicates a packet from an originating node (the originating node) to a destination node by separating the packet into multiple flow control digits (flits) and routing the flits using a series of hops over a set of intermediate nodes. The packet's final flit includes error detection information, such as checksum data. As each intermediate node receives the final flit, it performs error detection using the error detection information. The pattern of nodes that detect an error indicates which intermediate node has experienced a failure.
Claims10
A server system comprising:
- a network of nodes coupled in a network topology via a routing fabric, wherein:
- the nodes are to communicate a packet from an originating node to a destination node via a set of intermediate nodes;
- each intermediate node of the set of intermediate nodes is to perform error detection on the packet to produce a plurality of error reports, each of the plurality of error reports corresponding to a different one of the set of intermediate nodes;
- the server system is to identify a faulty node of the set of intermediate nodes based on the plurality of error reports;
- and
- the set of intermediate nodes form a routing path between the originating node and the destination node and the server system is to identify the faulty node by identifying the faulty node as a last intermediate node in the routing path whose corresponding error report does not indicate an error.
A method, comprising:
- communicating a packet from a first node of a server system to a second node of the server system via a set of intermediate nodes coupled to a routing fabric of the server system;
- performing error detection on the packet at each of the set of intermediate nodes to generate a plurality of error reports, each of the plurality of error reports corresponding to a different one of the set of intermediate nodes;
- and
- identifying at the server system a faulty node of the set of intermediate nodes based on the plurality of error reports;
- wherein the set of intermediate nodes form a routing path between the first node and the second node and identifying the faulty node corn rises identifying the faulty node as a last intermediate node in the routing path whose corresponding error report does not indicate an error.
Dependencies are inferred from the claim text.