Broadcom Corporation

National Semiconductor Corporation

Xilinx, Inc.

Skhynix Inc.

Cirrus Logic Inc.

Avago Technologies General IP (singapore) Pte. Ltd.

Avago Technologies International Sales Pte. Limited

Kenneth C Ma

Jeyhan Karaoguz

Wael William Diab

Bruce Joseph Currivan

Yongbum Kim

Michael David Johas Teener

Michael Johas Teener

Brima Babatunde Ibrahim

Sagheer Ahmad

Ygal Arbel

Balakrishna Jayadev

Dongwook Suh

Pranay D Shah

Vimi Pandey

Krishnan Srinivasan

John D Kenny

Scott Shaw

Yong Kim

Pete Edward Nelsen

Yossi Cohen

Seong-Ho Lee

Leo Jiang

Douglas Michael Berk

Millind Mittal

Michael D Johas-Teener

Kenneth J Kotlowski

Jeffrey A Hawkey

Some examples described herein provide for interconnect in chiplet systems, for example system-level techniques for error correction in chip-to-chip interfaces. In an example, a method of error correction includes receiving, at a first chiplet, a data message via a set of interconnect, and transmitting a first control message that requests retransmission of the data message based on detecting an error associated with receiving the data message. The method also includes transmitting one or more instances of a second control message that indicates an idle operation at the first chiplet until the first chiplet receives a third control message that triggers an end of a retransmission mode. The method also includes transmitting a fourth control message frame indicating the end of the retransmission mode, and receiving a retransmission of the data message from the second chiplet.

System-level techniques for error correction in chip-to-chip interfaces

Embodiments herein describe a multiple die system that includes an interposer that connects a first die to a second die. Each die has a bump interface structure that is connected to the other structure using traces in the interposer. However, the bump interface structures may have different orientations relative to each other, or one of the interface structures defines fewer signals than the other. Directly connecting the corresponding signals defined by the structures to each other may be impossible to do in the interposer, or make the interposer too costly. Instead, the embodiments here simplify routing in the interposer by connecting the signals in the bump interface structures in a way that simplifies the routing but jumbles the signals. The jumbled signals can then be corrected using reordering circuitry in the dies (e.g., in the link layer and physical layer).

Chip bump interface compatible with different orientations and types of devices

A memory-centric neural network system and operating method thereof includes: a processing unit; semiconductor memory devices coupled to the processing unit, the semiconductor memory devices containing instructions executed by the processing unit; a weight matrix constructed with rows and columns of memory cells, inputs of the memory cells of a same row being connected to one of axons, outputs of the memory cells of a same column being connected to one of neurons; timestamp registers registering timestamps of the axons and the neurons; and a lookup table containing adjusting values indexed in accordance with the timestamps, wherein the processing unit updates the weight matrix in accordance with the adjusting values.

Neural network hardware accelerator architectures and operating method thereof

Aspects of a method and system for content selection, delivery and payment may comprise a management entity that coordinates operation of one or more endpoint devices. The one or more endpoint devices may be operable to select one or more parameters associated with a particular content type from the management entity. The particular content type may specify a type of media content that is handled by the one or more endpoint devices. The one or more endpoint devices may be operable to receive configuration information from the management entity based on the selected one or more parameters associated with the particular content type. The one or more endpoint devices may be configured utilizing the received configuration information. The one or more endpoint devices may be operable to pay for the received configuration information.

Method and system for content selection, delivery and payment

One or more processors and/or one or more circuits may be operable to configure one or more virtual machines and a hypervisor for controlling the one or more virtual machines. The virtual machines and the hypervisor may be distributed across a plurality of network devices. A sub-hypervisor may be configured within each of the virtual machines utilizing the hypervisor. Load information of the network devices may be communicated to the hypervisor utilizing the sub-hypervisors. The virtual machines may include threads, may be load balanced utilizing the hypervisor, dynamically configured utilizing the hypervisor based on changes in the network devices, and scaled by the distribution of the virtual machines across the network devices. Information from the processing of data may be received in the virtual machines. The network devices may include a plurality of: servers, switches, routers, racks, blades, mainframes, personal data assistants, smart phones, desktop computers, and/or laptop devices.

Method and system for abstracting virtual machines in a network comprising plurality of hypervisor and sub-hypervisors

Communication devices may determine routes for packets based on packet marking, routing parameters and/or costs associated with routes. A route may be selected and the packets may be communicated via the selected route. The parameters may comprise service class, real time compression, packet preemption, quality measurements, tier bypass and/or power usage information. The costs may comprise capacity, efficiency and/or performance information for power usage, bandwidth, memory and/or processing. The marking may comprise traffic type, user device capabilities, service class, quality measurements, latency requirements and/or power usage information. Endpoint devices, software applications and/or service providers may insert the marking into packets. Routes may be determined and/or selected based on shortest path bridging, audio video bridging, the marking, the routing parameters and/or the costs. Parameters and/or costs may be received and/or discovered from communication devices. Packets and/or the marking may be parsed and/or inspected. Costs may be based on routing parameters.

Method and system for dynamic routing and/or switching in a network

A switching fabric, within a network switching device, may comprise a plurality of dedicated interface modules to enable the switching fabric to interface with a plurality of switching support subsystems. The switching support subsystems may comprise a memory subsystem, a content address memory (CAM) subsystem, a packet processing subsystem, and a physical layer (PHY) subsystem. The switching fabric may be operable to determine capacity and/or capability of each of the plurality of switching support subsystems; and to configure each of the plurality of dedicated interface modules based on the determined capacity and/or capability of a corresponding switching support subsystem. The switching fabric also comprises an inter-switch interface module to enable coupling the network switching device to a plurality of other network switching devices, to form a multi-device switching mesh that enable sharing of resources and/or aggregation of switching capabilities of the network switching devices.

Method and system for scalable switching architecture

Aspects of a method and system for predictive leasing of network services and applications based on a usage pattern may comprise a management entity that coordinates operation of one or more endpoint devices. A user profile associated with the one or more endpoint devices may be received by the management entity, wherein the management entity may utilize the user profile associated with the one or more endpoint devices and a usage pattern associated with the one or more endpoint devices to determine leasing information for the one or more endpoint devices. The management entity may be operable to communicate the determined leasing information to the one or more endpoint devices. The leasing information may comprise leasing services and/or applications to the one or more endpoint devices.

Method and system for leasing of network services and applications based on a usage pattern

Hypervisor functions, which may control operations of one or more virtual machines, may be distributed across a plurality of network devices. State information may be stored for the virtual machines on network devices for fault tolerance and resilience. The virtual machines may retrieve stored state information to recover from a fault. The hypervisor may control the storage of the state information. Resources of the network devices may be allocated for fault tolerance and resilience of the virtual machines based on network device parameters, which may include storage capacity, processor usage, access time, communication bandwidth, and/or latency. The state information may include program content, cache content, memory content, and/or register content information, may be stored on a continuous, periodic, or an aperiodic basis, and may be shared among the network devices to enable the processing of data by the virtual machines when a fault occurs, and may be incrementally updated.

Method and system for fault tolerance and resilience for virtualized machines in a network

Aspects of a method and system for customized data delivery and network configuration via aggregation of device attributes are provided. In this regard, a first network device may receive device attributes from a plurality of devices that are communicatively coupled to the first network device. The first network device may aggregate the received device attributes and communicate the aggregate attributes to a source network device. The first network device may receive customized content from the source network device, wherein the customization is based on the aggregate device attributes. The first network device may communicate the customized content to the plurality of devices. The first network device may comprise, for example, a set-top box, a wired access point, and/or wireless access point. The device attributes of a device may comprise settings, capabilities, and/or user preferences of the device.

Customized data delivery and network configuration via aggregation of device attributes

Aspects of a method and system for packetizing data for servicing traffic end-to-end are provided. In this regard, a networking subsystem may receive, from an application, one or more values of one or more parameters associated with data generated by the application. The networking subsystem may packetize the data into one or more packets. The networking subsystem may translate the received one or more values into one or more corresponding OSI layer 2 tags. The networking subsystem may tag the one or more packets with the corresponding OSI layer 2 tags and transmit the tagged packet(s) to one or more link partners. The link partners may thus process and forward the tagged packet(s) by inspecting only OSI layer 2 information of the tagged packet(s). The one or more corresponding OSI layer 2 tags may indicate, for example, quality of service required by, and/or protocols associated with, the data.

Method and system for packetizing data for servicing traffic end-to-end

Aspects of a method and system for network communications utilizing shared scalable resources are provided. In this regard, networking state information for one or more of a plurality of communication devices may be communicated to a network management device. The network management device may be operable to aggregate the networking state information. The plurality of communication devices may receive aggregated networking state information from the network management device. The plurality of communication devices may route packets based on the received aggregated networking state information. The network management device may be dynamically or manually selected from the plurality of communication devices. The plurality of communication devices may be associated with a sharing domain, and one or more communication devices may be dynamically added to and/or removed from the sharing domain.

Growing community of inventors

Cupertino, CA, United States of America

Kenneth C Ma