Mellanox Technologies, Ltd.

Roee Moyal

Ariel Shahar

Gil Bloch

Daniel Marcovitch

Yamin Friedman

Idan Burstein

Yuval Shpigelman

Dotan David Levi

Noam Bloch

Lior Narkis

Gal Yefet

Idan Borshteen

Richard Leigh Graham

Igor Voks

Eliel Peretz

Liran Liss

Natan Manevich

Ariel Almog

Boris Pismenny

Eyal Srebro

Dror Bohrer

Yossi Kuperman

Michael Kagan

Liron Mula

Paraskevas Bakopoulos

Diego Crupnicoff

Eitan Zahavi

Avi Urman

Wojciech Wasko

Adi Menachem

Alex Shpiner

Yossef Itigin

Ran Avraham Koren

Amit Mandelbaum

Dotan Finkelshtein

Kobi Pines

Dotan Finkelstein

Teferet Geula

Reut Zuck

Eran Ben Elisha

Richard Graham

Sean Pieper

Ali Ayoub

Gil Kremer

Shahrazad Hleihel

A computing device which may include a hardware-configurable device reconfigurable to perform a series of logical operations to determine, based on parameters related to execution of a job received by the hardware-configurable device, whether or not the job is permissible for execution.

Post-tapeout reconfigurable device

Technologies for payload direct memory storing (PDMS) for out-of-order delivery of packets in remote direct memory access (RDMA) are described. A responder device includes an RDMA transport layer that can receive packets out of order and allow direct data placement of packet data in order. The responder device receives a first packet with a first packet number and first location information. The responder device stores first packet data to a first location according to the first location information. The responder device also receives a second packet and stores second packet data to a second location according to the second location information. A second packet number indicates that the first packet is received out of order. The first and second packet data are stored in order. The responder device can provide an indication that a message has arrived in response to determining that all packets of the message have arrived.

Payload direct memory storing (PDMS) for remote direct memory access (RDMA)

Technologies for spreading packets of transport flows across multiple network paths are described. A network controller includes a transport layer and a network layer. The transport layer includes a flow scheduler to schedule a transport flow from one of a plurality of transport flows. The network layer includes multipath logic to receive packets from the transport flow and select which path of a plurality of paths to a destination to use for the packets based on path congestion weights corresponding to the plurality of paths.

Routing transport flows in a transport layer over multiple paths in a network layer

Technologies for spreading a burst of data across multiple network paths in remote direct memory access (RDMA) over converged Ethernet (RoCE) and InfiniBand are described. A network interface controller sends a first burst of a transport flow directed to a second node over a first network path. The network interface controller determines that a second burst is to be sent over a different network path, and identifies a second network path using a multipath context. The multipath context stores a first weight value or a first state associated with the first network path and a second weight value or a second state associated with the second network path. The network interface controller sends the second burst of data to the second node via the second network path.

Remote direct memory access (RDMA) multipath

A network adapter includes a network interface, a host interface and processing circuitry. The network interface connects to a communication network for communicating with remote targets. The host interface connects to a host that accesses a Multi-Channel Send Queue (MCSQ) storing Work Requests (WRs) originating from client processes running on the host. The processing circuitry is configured to retrieve WRs from the MCSQ and distribute the WRs among multiple Send Queues (SQs) accessible by the processing circuitry, and retrieve WRs from the multiple NSQs and execute data transmission operations specified in the WRs retrieved from the multiple NSQs.

Multi-channel connection management in a network adapter

Devices, methods, and systems are provided. In one example, a device is described to include a device interface that receives data from at least one data source; a data shuffle unit that collects the data received from the at least one data source, receives a descriptor that describes a data shuffle operation to perform on the data received from the at least one data source, performs the data shuffle operation on the collected data to produce shuffled data, and provides the shuffled data to at least one data target.

Data shuffle offload

In one embodiment, a system includes a memory to store a work queue including work queue entry slots, a processing device to write work queue entries to the work queue in a consecutive and cyclic manner, and a network device including a network interface to share packet over a network, and packet processing circuitry to read the work queue entries from the work queue in a consecutive and cyclic manner, the work queue entries indicating work to be performed associated with the packets, dequeue respective ones of the work queue entries read from the work queue responsively to reading the respective work queue entries from the work queue, add the work queue entries to an execution database used to track execution of the work queue entries, and execute the work queue entries in the execution database.

Efficient network device work queue

An apparatus includes a memory and control circuitry. The control circuitry is configured to receive packets, which are en-route to undergo transport-layer processing in a network device in accordance with a transport protocol that requires arrival of the packets in a sequential order, to detect that one or more of the packets deviate from the sequential order, to buffer the one or more deviating packets in the memory, and, using the memory, to reorder the packets and provide the packets in the sequential order to the network device.

Handling of out-of-order transport-layer packets using reorder buffer

A system includes a device coupled to a processing device. The processing device is to receive a request to execute a plurality of workloads, the request comprising a rate to execute each workload of the plurality of workloads and a parameter value indicating an execution offset. The processing device is further to determine a sequence for executing the plurality of workloads based on receiving the rate and the parameter value, where the sequence is to execute each workload at the respective rate and each workload of the plurality of workloads is executed at a different time based on the parameter value. The processing device is to execute the plurality of workloads in accordance with the sequence upon determining the sequence to execute the plurality of workloads.

Execution offset rate limiter

A system including a network interface layer, and a physical network connection configured to connect with a networking medium. The network interface layer is configured to: A) receive a plurality of user datagram protocol (UDP) message segments from the physical network connection; B) coalesce the plurality of UDP message segments into a coalesced UDP message; and C) send the coalesced UDP message to an application layer external to the system. Related apparatus and methods are also provided.

Message segmentation

A network adapter includes a network interface that communicates packets over a network, a host interface connected locally to a host processor and to a host memory, and processing circuitry, coupled between the network interface and the host interface, and is configured to receive in a common queue, via the host interface, (i) a processing work item specifying a source buffer in the host memory, a data processing operation, and a first address in the host memory, and (ii) an RDMA write work item specifying the first address, and a second address in a remote memory. In response to the processing work item, the processing circuitry reads data from the source buffer, applies the data processing operation, and stores the processed data in the first address. In response to the RDMA write work item the processing circuitry transmits the processed data, over the network, for storage in the second address.

Network adapter with a common queue for both networking and data manipulation work requests

Technologies for spreading a single transport flow across multiple network paths in remote direct memory access (RDMA) over converged Ethernet (RoCE) and InfiniBand are described. A network interface controller receives a first packet and a second packet of a transport flow directed to a second node. The network interface controller assigns a first network routing identifier to the first packet and a second network routing identifier to the second packet, the first network routing identifier corresponding to a first network path between the first and second nodes, the second network routing identifier corresponding to a second network path between the first node and the second node. The network interface controller schedules a first packet of the transport flow to be sent via the first network path and a second packet of the transport flow to be sent via the second network path.

A method for data transfer includes transmitting a sequence of data packets from a first computer over a network to a second computer in a single RDMA data transfer transaction. Upon receipt of a second packet in the sequence without previously having received the first packet, the second computer sends a NAK packet over the network to the first computer, indicating that the first packet was not received. A retransmission mode is selected responsively to the type of the transaction, such that when the transaction is of a first type, the first packet is retransmitted from the first computer to the second computer in response to the NAK packet without retransmitting the second packet, and when the transaction is of a second type, both the first and second packets are retransmitted from the first computer to the second computer in response to the NAK packet.

Selective retransmission of packets

A method using a memory and queue handling logic, including accessing a work control structure (WCS) configured 'first-in-first-out' holding work control records (WCRs) each including a field defining work to be carried out and a completion indicator indicating whether the work has completed, and initially set to indicate that the work has not completed: upon fetching a work request (WR) for execution, pushing a WCR corresponding to the WR to the WCS, and: A) inspecting the WCR at a head of the WCS, B) when the completion indicator of the WCR at the head of the WCS indicates that the unit of work associated with the WCR at the head of the WCS has been completed, popping the WCR at the head of the WCS from the WCS, and reporting completion of the WCR at the head of the WCS to a host processor, and C) iteratively performing A, B, and C.

Task completion system allowing tasks to be completed out of order while reporting completion in the original ordering my

Growing community of inventors

Yokneam Illit, Israel

Roee Moyal