Abstract: In an example, a system specifies a first configuration of the physical transport network that models a plurality of devices as a corresponding first plurality of nodes having a tree topology. Each node of the first plurality of nodes has at least one first device identifier and at least one first connection identifier to other nodes in the tree topology. The system specifies a second configuration of the logical transport network that models the plurality of devices as the first plurality of nodes having a non-tree topology. Each node of the first plurality of nodes has at least one second device identifier, at least one second connection identifier to other nodes in the non-tree topology, the at least one first device identifier, and the at least one first connection identifier of the tree topology.

Abstract: An example programmable device includes a configuration memory configured to store configuration data; a programmable logic having a configurable functionality based on the configuration data in the configuration memory; a signal conversion circuit; a digital processing circuit; an endpoint circuit coupled to the signal conversion circuit through the digital processing circuit; wherein the digital processing circuit includes a first one or more digital processing functions implemented as hardened circuits each having a predetermined functionality, and a second one or more processing functions implemented by the configurable functionality of the programmable logic.

Abstract: Embodiments herein describe transferring ownership of data (e.g., cachelines or blocks of data comprising multiple cachelines) from a host to hardware in an I/O device. In one embodiment, the host and I/O device (e.g., an accelerator) are part of a cache-coherent system where ownership of data can be transferred from a home agent (HA) in the host to a local HA in the I/O device—e.g., a computational slave agent (CSA). That way, a function on the I/O device (e.g., an accelerator function) can request data from the local HA without these requests having to be sent to the host HA. Further, the accelerator function can indicate whether the local HA tracks the data on a cacheline-basis or by a data block (e.g., multiple cachelines). This provides flexibility that can reduce overhead from tracking the data, depending on the function's desired use of the data.

Abstract: The present invention provides an antenna module for a massive MIMO antenna, the antenna module comprising a plurality of first signal ports, a number of first antenna elements arranged in a first matrix arrangement, wherein a number of rows of the first matrix arrangement and/or a number of columns of the first matrix arrangement equals the number of first signal ports, and a switching matrix that is configured to controllably couple each of the first signal ports either with all first antenna elements of a respective row of the first matrix arrangement or all first antenna elements of a respective column of the first matrix arrangement. Further, the present invention provides a respective massive MIMO antenna.

Abstract: Roughly described: a network interface device has an interface. The interface is coupled to first network interface device circuitry, host interface circuitry and host offload circuitry. The host interface circuitry is configured to interface to a host device and has a scheduler configured to schedule providing and/or receiving of data to/from the host device. The interface is configured to allow at least one of: data to be provided to said host interface circuitry from at least one of said first network device interface circuitry and said host offload circuitry; and data to be provided from said host interface circuitry to at least one of said first network interface device circuitry and said host offload circuitry.

Abstract: A method includes instantiating a simulation of an electronic design for a device under test (DUT) in hardware design language responsive to a user selection thereof. A subset of leaf nodes from a plurality of leaf nodes from the electronic design with input/output signaling of more than two values is identified. A hierarchical path for each leaf node of the plurality of leaf nodes of the electronic design for the DUT with respect to a testbench is calculated. A bypass module for the subset of leaf nodes is generated. The bypass module is generated in response to detecting presence of the subset of leaf nodes in the electronic design with input/output signaling of more than two values. The bypass module facilitates communication between the testbench and the subset of leaf nodes. Leaf nodes other than the subset of leaf nodes communicate with the testbench without communicating through the bypass module.

Abstract: Providing host-to-kernel streaming support can include determining a platform circuitry for use with a streaming kernel of a circuit design. The streaming kernel is configured for implementation in a user circuitry region of an integrated circuit (IC) to perform tasks offloaded from a host computer. The platform circuitry is configured for implementation in a static circuitry region of the IC. The platform circuitry is configured to establish a communication link with the host computer. An adaptable streaming controller can be inserted within the circuit design. The adaptable streaming controller is configured for implementation in the user circuitry region and connects to the streaming kernel. The adaptable streaming controller further communicatively links the streaming kernel with the platform circuitry. The adaptable streaming controller can be parameterized for exchanging data between the platform circuitry and the streaming kernel based, at least in part, on a type of the platform circuitry.

Abstract: A network interface device comprises a plurality of components configured to process a flow of data one after another. A control component is configured to provide one or more control messages in said flow, said one or more control message being provided to said plurality of components one after another such that a configuration of one or more of said components is changed.

Abstract: Embodiments herein describe circuitry with improved efficiency when executing layers in a nested neural network. As mentioned above, a nested neural network has at least one split operation where a tensor generated by a first layer is transmitted to, and processed by several branches in the neural network. Each of these branches can have several layers that have data dependencies which result in a multiply-add array sitting idly. In one embodiment, the circuitry can include a dedicated pre-pooler for performing a pre-pooling operation. Thus, the pre-pooling operation can be performing in parallel with other operations (e.g., the convolution performed by another layer). Once the multiply-add array is idle, the pre-pooling operation has already completed (or at least, has already started) which means the time the multiply-add array must wait before it can perform the next operation is reduced or eliminated.

Abstract: A network interface device having an FPGA for providing an FPGA application. A first interface between a host computing device and the FPGA application is provided, allowing the FPGA application to make use of data-path operations provided by a transport engine on the network interface device, as well as communicate with the host. The FPGA application sends and receives data with the host via a memory that is memory mapped to a shared memory location in the host computing device, whilst the transport engine sends and receives data packets with the host via a second memory. A second interface is provided to interface the FPGA application and transport engine with the network, wherein the second interface is configured to back-pressure the transport engine.

Abstract: Apparatus and associated methods relate to an ADC-based digital receiver including a feedforward equalizer (FFE) that has m precursor taps and n postcursor taps to equalize the precursor portion, and to adapt postcursor intersymbol interference (ISI) through a predetermined equalization coefficient selected to counteract the noise boosting effect associated with the precursor equalization. In an illustrative example, the receiver may dynamically balance noise and ISI through adaptively determining a coefficient hp1 of a first postcursor tap of a first FFE and a coefficient h1 of a first postcursor tap of a second equalizer adapted to substantially reduce or eliminate additional ISI introduced by the first FFE. The first FFE may optimize ISI removal and noise reduction, for example. One of the coefficients h1 and hp1 may be predetermined, and then the other coefficient may be iteratively adapted to trade off precursor ISI and postcursor ISI to minimize BER.

Abstract: A radome having an integrated antenna array and an antenna assembly having the same are described herein. A method for fabricating a radome having an integrated antenna array is also described herein. In one example, a radome is provided that includes a radome shell and an antenna array. The antenna array has a radiating surface and a backside surface. The radome shell is affixed to the antenna array forming an independent unitary structure separable from other components of an antenna assembly.

Abstract: Controlling execution of application-specific hardware pipelines includes detecting, using computer hardware, a loop construct contained in a function within a design specified in a high-level programming language, extracting, using the computer hardware, the loop construct from the function into a newly generated function of the design, and generating, using the computer hardware, a state transition graph corresponding to the loop construct. The state transition graph can be pruned by relocating operations from the function entry state and the function exit state into the loop region. A circuit design defining, at least in part, a pipeline hardware architecture implementing the loop construct can be generated using the computer hardware based, at least in part, on the pruned state transition graph.

Abstract: An integrated circuit having a data processing engine (DPE) array can include a plurality of memory tiles. A first memory tile can include a first direct memory access (DMA) engine, a first random-access memory (RAM) connected to the first DMA engine, and a first stream switch coupled to the first DMA engine. The first DMA engine may be coupled to a second RAM disposed in a second memory tile. The first stream switch may be coupled to a second stream switch disposed in the second memory tile.

Abstract: Some examples described herein provide for a shield in an integrated circuit (IC) structure for memory protection. In an example, an IC structure includes a semiconductor material, an interconnect structure, and a shield. The semiconductor material has a protected region. Devices are disposed in a first side of the semiconductor material in the protected region. The interconnect structure is disposed on the first side of the semiconductor material. The interconnect structure interconnects the devices in the protected region. The shield is disposed on a second side of the semiconductor material opposite from the first side of the semiconductor material. The shield is positioned aligned with the protected region.

Abstract: Changing accelerator card images without rebooting a host system includes receiving, within an integrated circuit (IC) of an accelerator card, an address of a platform image stored in a non-volatile memory of the accelerator card. The address is received over a communication link between the host system and the accelerator card while the communication link is connected. Changing accelerator card images includes detecting, within a register of the IC, that a warm boot enable flag is set and that the communication link with the host system is disconnected. In response to detecting that the warm boot enable flag is set and that the communication link is disconnected, loading of the platform image from the address of the non-volatile memory into the integrated circuit is initiated.

Abstract: A download dispatch circuit initiates download of an input tile of an input feature map in response to a source buffer of two or more source buffers being available for the input tile, and indicates that the input tile is available in response to completion of the download. An operation dispatch circuit initiates a neural network operation on the input tile in response to the input tile being available and a first destination buffer of two or more destination buffers being available for an output tile of an output feature map, and indicates that the output tile is available in response to completion of the neural network operation. An upload dispatch circuit initiates upload of the output tile to the output feature map in response to the output tile being available, and indicates that the first destination buffer is available in response to completion of the upload.

Abstract: A network interface device comprises an integrated circuit device comprises at least one processor. A network interface device comprises a memory. The integrated device is configured to execute a function with respect to at least a part of stored data in said memory.

Abstract: A semiconductor device comprises a plurality of memory elements, test control circuitry, and a testing interface. The test control circuitry is configure to determine that one or more clock signals associated with the memory elements have been stopped and generate a scan clock signal based on the determination that the one or more clock signals have been stopped. The test control circuitry is further configured to communicate the scan clock signal to the memory elements. The testing interface is configured to communicate test data from the memory elements. In one example, the test data is delimited with start and end marker elements. The semiconductor device is mounted to a circuit board and is communicatively coupled to communication pins of the circuit board.

Abstract: An integrated circuit can include a Network-on-Chip (NoC) having a router network with first and second shared physical channels. The NoC includes one or more master bridge circuits (MBCs) coupled to the router network, where each MBC provides a packet-based interface to a master client circuit coupled thereto for initiating transactions over the router network. Each MBC sends and receives data for the transactions over the router network as flits of packets according to a schedule. The NoC includes one or more slave bridge circuits (SBCs) coupled to the router network, where each SBC provides a packet-based interface to a slave client circuit coupled thereto to for responding to the transactions over the router network. Each SBC sends and receives the flits over the router network according to the schedule. The flits sent from different client circuits are interleaved using time-multiplexing on the first and second shared physical channels.