Abstract: A last-level collective hardware prefetcher (LLCHP) is described. The LLCHP is to detect a first off-chip memory access request by a first processor core of a plurality of processor cores. The LLCHP is further to determine, based on the first off-chip memory access request, that first data associated with the first off-chip memory access request is associated with second data of a second processor core of the plurality of processor cores. The LLCHP is further to prefetch the first data and the second data based on the determination.

Abstract: Disclosed herein are methods, systems, and devices for bandwidth steering. Systems may include a plurality of compute nodes configured to execute one or more applications, a plurality of first level resources communicatively coupled to the plurality of compute nodes, a plurality of second level resources communicatively coupled to the plurality of first level resources, and a plurality of third level resources communicatively coupled to the plurality of second level resources. Systems may also include a plurality of optical switch circuits communicatively coupled to the plurality of first level resources and the plurality of second level resources, wherein each of the plurality of optical switch circuits is coupled to more than one of the plurality of the first level resources and is also coupled to more than one of the plurality of the second level resources.

Abstract: A last-level collective hardware prefetcher (LLCHP) is described. The LLCHP is to detect a first off-chip memory access request by a first processor core of a plurality of processor cores. The LLCHP is further to determine, based on the first off-chip memory access request, that first data associated with the first off-chip memory access request is associated with second data of a second processor core of the plurality of processor cores. The LLCHP is further to prefetch the first data and the second data based on the determination.

Abstract: This disclosure provides systems, methods, and apparatus for collective memory transfers. A control unit may be configured to coordinate a transfer of data between a memory and processor cores. For a read data transfer operation, the control unit may receive a trigger packet identifying a read data transfer operation and identifying a first plurality of data lines based on data values included in the trigger packet. The control unit may read the first plurality of data lines from the memory sequentially and send a second plurality of data lines to the processor cores. For a write data transfer operation, the control unit may send a request for at least one data line to a plurality of processor cores, may receive and reorder the requested data lines, and may write the data lines to a memory. The control unit may determine a mapping between processor cores and the memory.

Abstract: This disclosure provides systems, methods, and apparatus for collective memory transfers. A control unit may be configured to coordinate a transfer of data between a memory and processor cores. For a read data transfer operation, the control unit may receive a trigger packet identifying a read data transfer operation and identifying a first plurality of data lines based on data values included in the trigger packet. The control unit may read the first plurality of data lines from the memory sequentially and send a second plurality of data lines to the processor cores. For a write data transfer operation, the control unit may send a request for at least one data line to a plurality of processor cores, may receive and reorder the requested data lines, and may write the data lines to a memory. The control unit may determine a mapping between processor cores and the memory.