Disclosed embodiments relate to instructions for vector operations with immediate values. In one example, a system includes a memory and a processor that includes fetch circuitry to fetch the instruction from a code storage, the instruction including an opcode, a destination identifier to specify a destination vector register, a first immediate, and a write mask identifier to specify a write mask register, the write mask register including at least one bit corresponding to each destination vector register element, the at least one bit to specify whether the destination vector register element is masked or unmasked, decode circuitry to decode the fetched instruction, and execution circuitry to execute the decoded instruction, to, use the write mask register to determine unmasked elements of the destination vector register, and, when the opcode specifies to broadcast, broadcast the first immediate to one or more unmasked vector elements of the destination vector register.

A processor includes: an address generating unit that, when an instruction decoded by a decoding unit is an instruction to execute arithmetic processing on a plurality of operand sets each including a plurality of operands that are objects of the arithmetic processing, in parallel a plurality of times, generates an address set corresponding to each of the operand sets of the arithmetic processing for each time, based on a certain address displacement with respect to the plurality of operands included in each of the operand sets; a plurality of instruction queues that hold the generated address sets corresponding to the respective operand sets, in correspondence to respective processing units; and a plurality of processing units that perform the arithmetic processing in parallel on the operand sets obtained based on the respective address sets outputted by the plurality of instruction queues.

A tensor traversal engine in a processor system comprising a source memory component and a destination memory component, the tensor traversal engine comprising: a control signal register storing a control signal for a strided data transfer operation from the source memory component to the destination memory component, the control signal comprising an initial source address, an initial destination address, a first source stride length in a first dimension, and a first source stride count in the first dimension; a source address register communicatively coupled to the control signal register; a destination address register communicatively coupled to the control signal register; a first source stride counter communicatively coupled to the control signal register; and control logic communicatively coupled to the control signal register, the source address register, and the first source stride counter.

Embodiments are generally directed to thread group scheduling for graphics processing. An embodiment of an apparatus includes a plurality of processors including a plurality of graphics processors to process data; a memory; and one or more caches for storage of data for the plurality of graphics processors, wherein the one or more processors are to schedule a plurality of groups of threads for processing by the plurality of graphics processors, the scheduling of the plurality of groups of threads including the plurality of processors to apply a bias for scheduling the plurality of groups of threads according to a cache locality for the one or more caches.

A method and device for providing a vector stream instruction set architecture extension for a CPU. In one aspect, there is provided a vector stream engine unit comprising: a first fast memory storage for temporarily storing data of vector data streams from a memory for loading into a vector register file; a second fast memory storage for temporarily storing data of the vector data streams from the vector register file for loading into the memory; a prefetcher configured to prefetch data of the vector data streams from the memory into the first fast storage memory, and prefetch data of the vector data streams from the vector register file into the second fast storage memory; and a stream configuration table (SCT) storing stream information for prefetching data from the vector data streams.

A tensor traversal engine in a processor system comprising a source memory component and a destination memory component, the tensor traversal engine comprising: a control signal register storing a control signal for a strided data transfer operation from the source memory component to the destination memory component, the control signal comprising an initial source address, an initial destination address, a first source stride length in a first dimension, and a first source stride count in the first dimension; a source address register communicatively coupled to the control signal register; a destination address register communicatively coupled to the control signal register; a first source stride counter communicatively coupled to the control signal register; and control logic communicatively coupled to the control signal register, the source address register, and the first source stride counter.

Systems, methods, and devices for pruning a convolutional neural network (CNN). A subset of layers of the CNN is chosen, and for each layer of the subset of layers, how salient each filter in the layer is to an output of the CNN is determined, a subset of the filters in the layer is determined based on the salience of each filter in the layer, and the subset of filters in the layer is pruned. In some implementations, the layers of the subset of layers of the CNN are non-contiguous. In some implementations, the subset of layers includes odd numbered layers of the CNN and excludes even numbered layers of the CNN. In some implementations, the subset of layers includes even numbered layers of the CNN and excludes odd numbered layers of the CNN.

A predicated-loop-terminating branch instruction controls, based on whether a loop termination condition is satisfied, whether the processing circuitry should process a further iteration of a predicated loop body or process a following instruction. If at least one unnecessary iteration of the predicated loop body is processed following a mispredicted-non-termination branch misprediction when the loop termination condition is mispredicted as unsatisfied for a given iteration when it should have been satisfied, processing of the at least one unnecessary iteration of the predicated loop body is predicated to suppress an effect of the at least one unnecessary iteration. When the mispredicted-non-termination branch misprediction is detected for the given iteration of the predicated-loop-terminating branch instruction, in response to determining that a flush suppressing condition is satisfied, flushing of the at least one unnecessary iteration of the predicated loop body is suppressed as a response to the mispredicted-non-termination branch misprediction.

In response to an instruction decoder decoding a range prefetch instruction specifying first and second address-range-specifying parameters and a stride parameter, prefetch circuitry controls, depending on the first and second address-range-specifying parameters and the stride parameter, prefetching of data from a plurality of specified ranges of addresses into the at least one cache. A start address and size of each specified range is dependent on the first and second address-range-specifying parameters. The stride parameter specifies an offset between start addresses of successive specified ranges. Use of the range prefetch instruction helps to improve programmability and improve the balance between prefetch coverage and circuit area of the prefetch circuitry.

An apparatus includes an array processor to process array data. The array data are arranged in a memory. The array data are specified with programmable per-dimension size and stride values.