Hewlett Packard Enterprise Development LP

Intel Corporation

Cray Incorporated

Robert L Alverson

Duncan Roweth

Steven Lee Scott

Keith D Underwood

Andrew Steven Kopser

Edwin Froese

Edwin L Froese

Dennis Charles Abts

Partha Kundu

Eric P Lundberg

Howard Pritchard

Jonathan P Beecroft

Joseph G Kopnick

Edward J Turner

Nathan L Wichmann

Michael A Parker

David Charles Hewson

Laurence S Kaplan

Albert S Cheng

Charles A Giefer

Timothy J Johnson

Christopher Michael Brueggen

Mike Parker

Steve Scott

Albert SauPong Cheng

Igor Gorodetsky

Burton Jordan Smith

Abdulla M Bataineh

Gail A Alverson

Thomas L Court

James Robert Kohn

Anthony M Ford

Matthew Paul Kelly

Gregory J Faanes

Abdulla Bataineh

Trevor Alan Jones

Gregory Faanes

Michael Higgins

Steven L Scott

Skef Iterum

James Reinhard

Konstantinos Fragkiadakis

Daniel C Duval

Eric A Hoffman

Kazuya Okubo

Mark Swan

Asaph Zemach

Systems and methods are provided for providing a shared token as part of establishing a communication connection with a network ('connection establishment') between a source device and a target device, after a previous communication connection to the network between the source device and the target device is closed or torn down, in helping to prevent future replay attacks.

Secure connection initiation to prevent replay attacks in an encrypted network

Systems and methods are provided for passing data amongst a plurality of switches having a plurality of links attached between the plurality of switches. At a switch, a plurality of load signals are received from a plurality of neighboring switches. Each of the plurality of load signals are made up of a set of values indicative of a load at each of the plurality of neighboring switches providing the load signal. Each value within the set of values provides an indication for each link of the plurality of links attached thereto as to whether the link is busy or quiet. Based upon the plurality of load signals, an output link for routing a received packet is selected, and the received packet is routed via the selected output link.

Algorithms for use of load information from neighboring nodes in adaptive routing

An apparatus is provided that includes a network interface to transmit and receive data packets over a network; a memory including one or more buffers; an arithmetic logic unit to perform arithmetic operations for organizing and combining the data packets; and a circuitry to receive, via the network interface, data packets from the network; aggregate, via the arithmetic logic unit, the received data packets in the one or more buffers at a network rate; and transmit, via the network interface, the aggregated data packets to one or more compute nodes in the network, thereby optimizing latency incurred in combining the received data packets and transmitting the aggregated data packets, and hence accelerating a bulk data allreduce operation. One embodiment provides a system and method for performing the allreduce operation. During operation, the system performs the allreduce operation by pacing network operations for enhancing performance of the allreduce operation.

System and method for implementing a network-interface-based allreduce operation

A network interface controller (NIC) capable of performing message passing interface (MPI) list matching is provided. The NIC can include a host interface, a network interface, and a hardware list-processing engine (LPE). The host interface can couple the NIC to a host device. The network interface can couple the NIC to a network. During operation, the LPE can receive a match request and perform MPI list matching based on the received match request.

System and method for facilitating efficient message matching in a network interface controller (NIC)

A network interface card (NIC) receives a stream of commands, a respective command comprising memory-operation requests, each request associated with a destination NIC. The NIC buffers asynchronously the requests into queues based on the destination NIC, each queue specific to a corresponding destination NIC. When first queue requests reach a threshold, the NIC aggregates the first queue requests into a first packet and sends the first packet to the destination NIC. The NIC receives a plurality of packets, a second packet comprising memory-operation requests, each request associated with a same destination NIC and a destination core. The NIC buffers asynchronously the requests of the second packet into queues based on the destination core, each queue specific to a corresponding destination core. When second queue requests reach the threshold, the NIC aggregates the second queue requests into a third packet and sends the third packet to the destination core.

Aggregating small remote memory access requests

A switch equipped with a self-managing reduction engine is provided. During operation, the reduction engine can use a timeout mechanism to manage itself in different latency-induced or error scenarios. As a result, the network can facilitate an efficient and scalable environment for high performance computing.

System and method for facilitating self-managing reduction engines

A switch capable of on-the-fly reduction in a network is provided. The switch is equipped with a reduction engine that can be dynamically configured to perform on-the-fly reduction. As a result, the network can facilitate an efficient and scalable environment for high performance computing.

System and method for performing on-the-fly reduction in a network

A network interface card (NIC) receives packets corresponding to a read or write request, the packets associated with a datatype descriptor stored in a datatype engine of the NIC, and each packet associated with a precomputed context which indicates a value for each dimension of a multi-dimensional array and a start location of the respective packet within a host memory block. The NIC generates, for a respective packet, a datatype handle corresponding to the datatype descriptor and an offset indicating a position of the respective packet within the packets. The NIC determines, based on the datatype handle and the offset, a cached context for the respective packet and initializes the datatype engine based on the cached context. The datatype engine generates, based on the cached context, read or write requests comprising addresses and lengths, thereby allowing the NIC to process out-of-order packets based on the precomputed and cached context.

Mechanism to enable out-of-order packet processing in a datatype engine

A network interface controller (NIC) capable of efficient packet forwarding is provided. The NIC can be equipped with a host interface, a packet generation logic block, and a forwarding logic block. During operation, the packet generation logic block can obtain, via the host interface, a message from the host device and for a remote device. The packet generation logic block may generate a plurality of packets for the remote device from the message. The forwarding logic block can then send a first subset of packets of the plurality of packets based on ordered delivery. If a first condition is met, the forwarding logic block can send a second subset of packets of the plurality of packets based on unordered delivery. Furthermore, if a second condition is met, the forwarding logic block can send a third subset of packets of the plurality of packets based on ordered delivery.

System and method for facilitating efficient packet forwarding in a network interface controller (NIC)

A method and apparatus are provided for facilitating a datatype engine (DTE) to support high performance computing. A network interface card (NIC) receives, via a message passing interface, a command to read data from a host memory. The NIC determines that the command indicates a first datatype descriptor stored in the NIC. The NIC forms, based on the command, a packet which indicates a base address and a length associated with the data to be read from the host memory and passes the packet to the DTE. The DTE generates a plurality of read requests comprising offsets from the base address and corresponding lengths based on the first datatype descriptor. The DTE passes the plurality of read requests to a direct memory access module, thereby allowing the NIC to access the host memory while eliminating copies of the data on the host during transfer of the command across a network.

Datatype engine to support high performance computing

A system, method, and apparatus are provided to facilitate data structures for a datatype engine and provide inline compaction. The system receives, by a network interface card (NIC), a command to read data from a host memory, wherein the command indicates a datatype. The system generates a plurality of read requests comprising offsets from a base address and corresponding lengths based on the datatype. The system issues the plurality of read requests to the host memory to obtain the data from the host memory. The system obtains a byte-mask descriptor corresponding to the datatype. The system performs, based on the obtained data and the byte-mask descriptor, on-the-fly compaction of the obtained data, thereby allowing the NIC to return a requested subset of the obtained data.

Mechanism to represent data structures for a datatype engine and provide inline compaction

Systems and methods of routing a data communication across a network having a plurality switches are provided by monitoring the operation of the plurality of global links to determine which of the plurality of global links provide working paths. A routing table indicative of a status for the plurality of links is maintained, where the routing table provides weighting for each of the working paths. When routing, a link using a weighted pseudo-random selection from the choices available in the routing table is selected. Routing along one of the working paths commensurate with the selected link is performed, and the weighting is updated based upon the operation of the plurality of links.

Weighting routing

Systems and methods are provided for managing multicast data transmission in a network having a plurality of switches arranged in a Dragonfly network topology, including: receiving a multicast transmission at an edge port of a switch and identifying the transmission as a network multicast transmission; creating an entry in a multicast table within the switch; routing the multicast transmission across the network to a plurality of destinations via a plurality of links, wherein at each of the links the multicast table is referenced to determine to which ports the multicast transmission should be forwarded; and changing, when necessary, the virtual channel used by each copy of the multicast transmission as the copy progresses through the network.

Deadlock-free multicast routing on a dragonfly network

Growing community of inventors

Seattle, WA, United States of America

Robert L Alverson