An accelerator includes: a memory configured to store input data; a plurality of shift buffers each configured to shift input data received sequentially from the memory in each cycle, and in response to input data being stored in each of internal elements of the shift buffer, output the stored input data to a processing element (PE) array; a plurality of backup buffers each configured to store input data received sequentially from the memory and transfer the stored input data to one of the shift buffers; and the PE array configured to perform an operation on input data received from one or more of the shift buffers and on a corresponding kernel.

Distributed processors and methods for compiling code for execution by distributed processors are disclosed. In one implementation, a distributed processor may include a substrate; a memory array disposed on the substrate; and a processing array disposed on the substrate. The memory array may include a plurality of discrete memory banks, and the processing array may include a plurality of processor subunits, each one of the processor subunits being associated with a corresponding, dedicated one of the plurality of discrete memory banks. The distributed processor may further include a first plurality of buses, each connecting one of the plurality of processor subunits to its corresponding, dedicated memory bank, and a second plurality of buses, each connecting one of the plurality of processor subunits to another of the plurality of processor subunits.

Distributed processors and methods for compiling code for execution by distributed processors are disclosed. In one implementation, a distributed processor may include a substrate; a memory array disposed on the substrate; and a processing array disposed on the substrate. The memory array may include a plurality of discrete memory banks, and the processing array may include a plurality of processor subunits, each one of the processor subunits being associated with a corresponding, dedicated one of the plurality of discrete memory banks. The distributed processor may further include a first plurality of buses, each connecting one of the plurality of processor subunits to its corresponding, dedicated memory bank, and a second plurality of buses, each connecting one of the plurality of processor subunits to another of the plurality of processor subunits.

An apparatus and method for intelligently scheduling threads across a plurality of logical processors. For example, one embodiment of a processor comprises: a plurality of cores to be allocated to form a first plurality of logical processors (LPs) to execute threads, wherein one or more logical processors (LPs) are to be associated with each core of the plurality of cores; scheduling guide circuitry to: monitor execution characteristics of the first plurality of LPs and the threads; generate a first plurality of LP rankings, each LP ranking including all or a subset of the plurality of LPs in a particular order; and store the first plurality of LP rankings in a memory to be provided to a scheduler, the scheduler to schedule the threads on the plurality of LPs using the first plurality of LP rankings; a power controller to execute power management code to perform power management operations including independently adjusting frequencies and/or voltages of one or more of the plurality of cores; wherein in response to a core configuration command to deactivate a first core of the plurality of cores, the power controller or privileged program code executed on the processor are to update the memory with an indication of deactivation of the first core, wherein responsive to the indication of deactivation of the first core, the scheduler is to modify the scheduling of the threads.

An accelerator includes: a memory configured to store input data; a plurality of shift buffers each configured to shift input data received sequentially from the memory in each cycle, and in response to input data being stored in each of internal elements of the shift buffer, output the stored input data to a processing element (PE) array; a plurality of backup buffers each configured to store input data received sequentially from the memory and transfer the stored input data to one of the shift buffers; and the PE array configured to perform an operation on input data received from one or more of the shift buffers and on a corresponding kernel.

Distributed processors and methods for compiling code for execution by distributed processors are disclosed. In one implementation, a distributed processor may include a substrate; a memory array disposed on the substrate; and a processing array disposed on the substrate. The memory array may include a plurality of discrete memory banks, and the processing array may include a plurality of processor subunits, each one of the processor subunits being associated with a corresponding, dedicated one of the plurality of discrete memory banks. The distributed processor may further include a first plurality of buses, each connecting one of the plurality of processor subunits to its corresponding, dedicated memory bank, and a second plurality of buses, each connecting one of the plurality of processor subunits to another of the plurality of processor subunits.

Embodiments of a multi-processor array are disclosed that may include a plurality of processors, local memories, configurable communication elements, and direct memory access (DMA) engines, and a DMA controller. Each processor may be coupled to one of the local memories, and the plurality of processors, local memories, and configurable communication elements may be coupled together in an interspersed arrangement. The DMA controller may be configured to control the operation of the plurality of DMA engines.

Distributed processors and methods for compiling code for execution by distributed processors are disclosed. In one implementation, a distributed processor may include a substrate; a memory array disposed on the substrate; and a processing array disposed on the substrate. The memory array may include a plurality of discrete memory banks, and the processing array may include a plurality of processor subunits, each one of the processor subunits being associated with a corresponding, dedicated one of the plurality of discrete memory banks. The distributed processor may further include a first plurality of buses, each connecting one of the plurality of processor subunits to its corresponding, dedicated memory bank, and a second plurality of buses, each connecting one of the plurality of processor subunits to another of the plurality of processor subunits.

A mainboard and a server are provided. The mainboard includes: a board body, a preset number of Purley platform central processors, and one or more memories. The preset number of Purley platform central processors and the one or more memories are installed on the board body. The Purley platform central processors are sequentially connected with each other, and each of the memories is connected to one of the Purley platform central processors. Each of the memories is configured to receive to-be-burned data inputted from outside and transmit the to-be-burned data to the Purley platform central processor connected with the memory. Each of the Purley platform central processors is configured to burn the to-be-burned data when receiving the to-be-burned data transmitted by the connected memory connected with the Purley platform central processor, to have a function corresponding to the to-be-burned data.

An apparatus and method for intelligently scheduling threads across a plurality of logical processors. For example, one embodiment of a processor comprises: a plurality of cores to be allocated to form a first plurality of logical processors (LPs) to execute threads, wherein one or more logical processors (LPs) are to be associated with each core of the plurality of cores; scheduling guide circuitry to: monitor execution characteristics of the first plurality of LPs and the threads; generate a first plurality of LP rankings, each LP ranking including all or a subset of the plurality of LPs in a particular order; and store the first plurality of LP rankings in a memory to be provided to a scheduler, the scheduler to schedule the threads on the plurality of LPs using the first plurality of LP rankings; a power controller to execute power management code to perform power management operations including independently adjusting frequencies and/or voltages of one or more of the plurality of cores; wherein in response to a core configuration command to deactivate a first core of the plurality of cores, the power controller or privileged program code executed on the processor are to update the memory with an indication of deactivation of the first core, wherein responsive to the indication of deactivation of the first core, the scheduler is to modify the scheduling of the threads.