Abstract: The present disclosure advantageously provides a matrix expansion unit that includes an input data selector, a first register set, a second register set, and an output data selector. The input data selector is configured to receive first matrix data in a columnwise format. The first register set is coupled to the input data selector, and includes a plurality of data selectors and a plurality of registers arranged in a first shift loop. The second register set is coupled to the data selector, and includes a plurality of data selectors and a plurality of registers arranged in a second shift loop. The output data selector is coupled to the first register set and the second register set, and is configured to output second matrix data in a rowwise format.

Abstract: A multi-thread processor has a canonical thread map register which outputs a sequence of thread_id values indicating a current thread for execution. The thread map register is programmable to provide granularity of number of cycles of the canonical sequence assigned to each thread. In one example of the invention, the thread map register has repeating thread identifiers in a sequential or non-sequential manner to overcome memory latency and avoid thread stalls. In another example of the invention, separate interrupt tasks are placed on each thread to reduce interrupt processing latency.

Abstract: A discrete three-dimensional (3-D) processor comprises communicatively coupled first and second dice. The first die comprises 3-D memory (3D-M) arrays, whereas the second die comprises at least a non-memory circuit and at least an off-die peripheral-circuit component of the 3D-M arrays. The first die does not comprise said off-die peripheral-circuit component. The non-memory circuit on the second die is not part of a memory.

Abstract: A method includes setting an output of each memory cell in an array of memory cells to a same first value, decreasing power to the array of memory cells and then increasing power to the array of memory cells. Memory cells in the array of memory cells with outputs that switched to a second value different from the first value are then identified in response to decreasing and then increasing the power. A set of memory cells is then selected from the identified memory cells to use in hardware security.

Abstract: An input/output store instruction is handled. A data processing system includes a system nest communicatively coupled to at least one input/output bus by an input/output bus controller. The data processing system further includes at least a data processing unit including a core, system firmware and an asynchronous core-nest interface. The data processing unit is communicatively coupled to the system nest via an aggregation buffer. The system nest is configured to asynchronously load from and/or store data to an external device which is communicatively coupled to the input/output bus. The data processing unit is configured to complete the input/output store instruction before an execution of the input/output store instruction in the system nest is completed.

Abstract: The disclosure relates to an artificial intelligence (AI) system that simulates functions such as cognition and judgment of the human brain by utilizing machine learning algorithms such as deep learning and its applications.

Abstract: A discrete three-dimensional (3-D) processor comprises first and second dice. The first die comprises 3-D memory (3D-M) arrays and in-die peripheral-circuit components thereof, whereas the second die comprises processing circuits and off-die peripheral-circuit components of the 3D-M arrays. The first and second dice are communicatively coupled by a plurality of inter-die connections.

Abstract: A discrete three-dimensional (3-D) processor comprises first and second dice. The first die comprises 3-D random-access memory or 3-D read-only memory (3D-RAM/3D-ROM) arrays, whereas the second die comprises logic circuits and at least an off-die peripheral-circuit component of the 3D-RAM/3D-ROM arrays. The first die does not comprise the off-die peripheral-circuit component of the 3D-RAM/3D-ROM arrays.

Abstract: An instruction processing apparatus is disclosed and includes: an instruction cache, which maps data blocks in a memory based on a multi-way set-associative structure and includes a plurality of cache lines; and an access control unit, coupled between an instruction fetch unit and the instruction cache, and adapted to read the plurality of cache lines respectively by using a plurality of data channels, and select a hit cache line from the plurality of cache lines by using a plurality of selection channels, to obtain an instruction, where the access control unit includes a path prediction unit, where the path prediction unit obtains, based on a type of the instruction, path prediction information corresponding to an instruction address, and enables at least one data channel and/or at least one selection channel based on the path prediction information.

Abstract: An RISC-V-based 3D interconnected multi-core processor architecture and a working method thereof. The RISC-V-based 3D interconnected multi-core processor architecture includes a main control layer, a micro core array layer and an accelerator layer, wherein the main control layer includes a plurality of main cores which are RISC-V instruction set CPU cores, the micro core array layer includes a plurality of micro unit groups including a micro core, a data storage unit, an instruction storage unit and a linking controller, wherein the micro core is an RISC-V instruction set CPU core that executes partial functions of the main core; the accelerator layer is configured to optimize a running speed of space utilization for accelerators meeting specific requirements, wherein some main cores in the main control layer perform data interaction with the accelerator layer, the other main cores interact with the micro core array layer.

Abstract: A system includes a memory-mapped register (MMR) associated with a claim logic circuit, a claim field for the MMR, a first firewall for a first address region, and a second firewall for a second address region. The MMR is associated with an address in the first address region and an address in the second address region. The first firewall is configured to pass a first write request for an address in the first address region to the claim logic circuit associated with the MMR. The claim logic circuit associated with the MMR is configured to grant or deny the first write request based on the claim field for the MMR. Further, the second firewall is configured to receive a second write request for an address in the second address region and grant or deny the second write request based on a permission level associated with the second write request.

Abstract: A system and method of processing instructions may comprise an application processing domain (APD) and a metadata processing domain (MTD). The APD may comprise an application processor executing instructions and providing related information to the MTD. The MTD may comprise a tag processing unit (TPU) having a cache of policy-based rules enforced by the MTD. The TPU may determine, based on policies being enforced and metadata tags and operands associated with the instructions, that the instructions are allowed to execute (i.e., are valid). The TPU may write, if the instructions are valid, the metadata tags to a queue. The queue may (i) receive operation output information from the application processing domain, (ii) receive, from the TPU, the metadata tags, (iii) output, responsive to receiving the metadata tags, resulting information indicative of the operation output information and the metadata tags; and (iv) permit the resulting information to be written to memory.

Abstract: A hardware neural network engine which uses checksums of the matrices used to perform the neural network computations. For fault correction, expected checksums are compared with checksums computed from the matrix developed from the matrix operation. The expected checksums are developed from the prior stage of the matrix operations or from the prior stage of the matrix operations combined with the input matrices to a matrix operation. This use of checksums allows reading of the matrices from memory, the dot product of the matrices and the accumulation of the matrices to be fault corrected without triplication of the matrix operation hardware and extensive use of error correcting codes. The nonlinear stage of the neural network computation is done using triplicated nonlinear computational logic. Fault detection is done in a similar manner, with fewer checksums needed and correction logic removed as compared to the fault correction operation.

Abstract: A discrete three-dimensional (3-D) processor comprises first and second dice. The first die comprises 3-D memory (3D-M) arrays, whereas the second die comprises logic circuits and at least an off-die peripheral-circuit component of the 3D-M array(s). Typical off-die peripheral-circuit component could be an address decoder, a sense amplifier, a programming circuit, a read-voltage generator, a write-voltage generator, a data buffer, or a portion thereof.

Abstract: Technologies for hybrid field-programmable gate array (FPGA) application-specific integrated circuit (ASIC) code acceleration are described. In one example, the computing device includes a FPGA comprising: algorithm circuitry to: perform one or more algorithm tasks of an algorithm, wherein the algorithm to perform a service request that is offloaded to the FPGA; and determine a primitive task associated with an algorithm task of the one or more algorithm tasks; primitive offload circuitry to encapsulate the primitive task in a buffer of the FPGA, wherein the buffer is accessible by an ASIC of the computing device; and result circuitry to return one or more results of the service request responsive to performance of the primitive task by the ASIC.

Abstract: Embodiments of a device include an integrated circuit, a reconfigurable stream switch formed in the integrated circuit along with a plurality of convolution accelerators and an arithmetic unit coupled to the reconfigurable stream switch. The arithmetic unit has at least one input and at least one output. The at least one input is arranged to receive streaming data passed through the reconfigurable stream switch, and the at least one output is arranged to stream resultant data through the reconfigurable stream switch. The arithmetic unit also has a plurality of data paths. At least one of the plurality of data paths is solely dedicated to performance of operations that accelerate an activation function represented in the form of a piece-wise second order polynomial approximation.

Abstract: An extensible multi-precision data pipeline system, comprising, a local buffer that stores an input local data set in a local storage format, an input tensor shaper coupled to the local buffer that reads the input local data set and converts the input local data set into an input tensor data set having a tensor format of vector width N by tensor length L, a cascaded pipeline coupled to the input tensor shaper that routes the input tensor data set through at least one function stage resulting in an output tensor data set, an output tensor shaper coupled to the cascaded pipeline that converts the output tensor data set into an output local data set having the local storage format and wherein the output tensor shaper writes the output local data set to the local buffer.

Abstract: A system for PTO-driven refrigeration includes a generator that is configured to be mechanically connected to a power takeoff (PTO) and a converter that is configured to receive AC power from the generator and is operable to convert the AC power to DC power. The generator is connected to a charge controller that is connected to an energy storage element. The energy storage element is connected to a controller configured to receive DC power and provide AC power to a motor. The motor may be mechanically connectable to a refrigeration system. The energy storage element is further configured to receive power from a second charge controller that receives power via an AC power input or solar system.

Abstract: A computer-controlled motorized pump system is provided. A generator is mechanically connected to a power takeoff. A first controller receives AC power from the generator and converts the AC power to DC power and provides DC power to a computing system that has one or more processors and one or more computer-readable hardware storage media and a user interface. A second controller is directly coupled to the first controller and provides AC power to a motor. The motor is mechanically connected to a pump, and the motor is in communication with, or controlled by, the computing system.

Abstract: Described herein are various embodiments of reducing dynamic power consumption within a processor device. One embodiment provides a technique for dynamic link width reduction based on throughput demand for client of an interconnect fabric. One embodiment provides for a parallel processor comprising an interconnect fabric including a dynamically configurable bus widths and frequencies.