Remote memory operations
AssigneeIntel Corporation
InventorsFrancesc GUIM BERNAT, Karthik KUMAR, Thomas WILLHALM, Narayan RANGANATHAN, Pete D. VOGT
Life of this patent
Estimating the term…
Abstract
An extension of node architecture and proxy requests enables a node to expose memory computation capability to remote nodes. A remote node can request execution of an operation by a remote memory computation resource, and the remote memory computation resource can execute the request locally and return the results of the computation. The node includes processing resources, a fabric interface, and a memory subsystem including a memory computation resource. The local execution of the request by the memory computation resource can reduce latency and bandwidth concerns typical with remote requests.
Claims23
A node in a network of nodes, comprising:
- a fabric interface to couple to the network of nodes;
- a processor to execute operations at the node;
- and
- a memory subsystem with a memory computation resource separate from the processor;
- wherein the memory subsystem to receive a request from a remote node processor for a memory computation, and in response to the request, to perform the computation locally at the memory subsystem and send the result of the computation to the remote node.
A network, comprising:
- an interconnection fabric;
- and
- multiple nodes coupled via the interconnection fabric, wherein a node including
- a fabric interface to couple to the interconnection fabric;
- a processor to execute operations at the node;
- and
- a memory subsystem with a memory computation resource separate from the processor;
- wherein a first node is to provide a request for a memory computation to a second node to cause the memory subsystem at the second node to perform the computation locally at the memory subsystem of the second node and send the result of the computation to the first node.
A method of operation in a network of nodes, comprising:
- receiving at a target node a request for a memory computation from a remote node over an interconnection fabric;
- responsive to receiving the request, performing the memory computation locally at a memory subsystem of the target node;
- and
- sending a result of the computation to the remote node.
Dependencies are inferred from the claim text.