Abstract: In some embodiments, multiple functional blocks (agents) in a complex integrated circuit are connected to a physically-distant shared resource (e.g. a memory controller) through packet buses which do not depend on establishing clock-cycle sequenced handshakes. On-chip repeaters including one or more register stages are used to segment the agent-shared resource interconnects into multiple segments, each shorter than a single-clock-cycle pathlength. The interconnects of multiple closely-spaced agents can be routed to the shared resource through an on-chip router having a single routed connection to the shared resource, for reducing the floorplan space taken by interconnects. The packet-based communications protocols do not require redesigning the agents or memory controller to make protocol changes accounting for the clock cycles inserted by repeaters and/or routers.

Abstract: In some embodiments, the execution of load and store instructions for internal fields of data structures is accelerated by using on-chip template registers and appropriate machine code instructions. A load/store machine code instruction comprises an identifier of a memory address offset of an internal field word relative to a base address of the data structure, an identifier of an intra-word start bit of the internal field, and an identifier of an intra-word length of the internal field. The three identifiers may coincide, for example if the three identifiers are represented by an identity of a template register storing a template entry including the memory address offset, the start position, and the field length. The three identifiers may also be provided as part of a machine code instruction itself. Further provided are compilers, compiler methods, and hardware systems for implementing accelerated internal-field load and store operations.

Abstract: A data processing system architecture is based upon a hardware engine that includes a plurality of functional units and data routing units that interconnect the functional units. The hardware engine performs operations and computations on data as the data traverses paths through the functional units under control of software. The functional units include logic resources, examples of which are flip-flops, latches, arithmetic logic units, random access memory, and the like. The routing units are responsive to the software control signals that are turned on or off to steer the data through these resources. Operations and computations are accomplished according to the steering of the data through the functional units that control the functions performed.

Abstract: A dual-mode network storage controller integrated on a chip is connected to a first set of hosts over a block-level storage area network (SAN), and to a second set of hosts over a metadata Ethernet/IP network. The dual-mode storage controller is also connected to one or more storage devices, such as a Redundant Array of Independent Disks (RAID). The storage controller comprises dedicated-hardware metadata translation logic for translating metadata (e.g. file-level) storage commands into block-level storage commands. The storage controller can also include block translation logic for translating logical block-level storage commands into physical block-level storage commands. The storage controller further comprises multiplexing logic for sequentially transmitting to the storage device(s) block level storage commands derived from the commands received from the first set of hosts and the second set of hosts.