An apparatus and method for performing a vector packed multiplication of signed and unsigned words. For example, one embodiment of a processor includes a decoder to decode a vector packed multiply instruction having operands to identify a first and a second plurality of packed words, first and second source registers to store the first and second plurality of packed words, and execution circuitry to execute the decoded instruction. The execution circuitry includes multiplier circuitry to multiply each packed word in the first source register with a corresponding packed word in the second source register to generate a plurality of doubleword products and rounding circuitry to round each of the doubleword products according to a rounding method to generate a plurality of rounded doubleword products. Each upper word of the rounded doubleword results is then stored into a corresponding word data element positions of a destination register.

Examples of a carry chain for performing an operation on operands each including elements of a selectable size is provided. Advantageously, the carry chain adapts to elements of different sizes. The carry chain determines a mask based on a selected size of an element. The carry chain selects, based on the mask, whether to carry a partial result of an operation performed on corresponding first portions of a first operand and a second operand into a next operation. The next operation is performed on corresponding second portions of the first operand and the second operand, and, based on the selection, the partial result of the operation. The carry chain stores, in a memory, a result formed from outputs of the operation and the next operation.

A data processing apparatus, a method of operating a data processing apparatus, a non-transitory computer readable storage medium, and an instruction are provided. The instruction specifies a first source register and a second source register. In response to the instruction control signals are generated, causing processing circuitry to perform a dot product operation. For this operation at least a first data element and a second data element are extracted from each of the first source register and the second source register, such that then at least first data element pairs and second data element pairs are multiplied together. The dot product operation is performed independently in each of multiple intra-register lanes across each of the first source register and the second source register. A widening operation with a large density of operations per instruction is thus provided.

A pipeline includes a first portion configured to process a first subset of bits of an instruction and a second portion configured to process a second subset of the bits of the instruction. A first clock mesh is configured to provide a first clock signal to the first portion of the pipeline. A second clock mesh is configured to provide a second clock signal to the second portion of the pipeline. The first and second clock meshes selectively provide the first and second clock signals based on characteristics of in-flight instructions that have been dispatched to the pipeline but not yet retired. In some cases, a physical register file is configured to store values of bits representative of instructions. Only the first subset is stored in the physical register file in response to the value of the zero high bit indicating that the second subset is equal to zero.

An apparatus and method to execute ray tracing instructions. For example, one embodiment of an apparatus comprises execution circuitry to execute a dequantize instruction to convert a plurality of quantized data values to a plurality of dequantized data values, the dequantize instruction including a first source operand to identify a plurality of packed quantized data values in a source register and a destination operand to identify a destination register in which to store a plurality of packed dequantized data values, wherein the execution circuitry is to convert each packed quantized data value in the source register to a floating point value, to multiply the floating point value by a first value to generate a first product and to add the first product to a second value to generate a dequantized data value, and to store the dequantized data value in a packed data element location in the destination register.

An information processing method includes acquiring first output data for input data of first learning model, reference data for the input data, and second output data for the input data of second learning model obtained by converting first learning model; calculating first difference data corresponding to a difference between the first difference data and the reference data and second difference data corresponding to a difference between the second output data and the reference data; and training first learning model with use of the first difference data and the second difference data.

Systems, apparatuses, and methods related to bit string operations in memory are described. The bit string operations may be performed within a memory array without transferring the bit strings or intermediate results of the operations to circuitry external to the memory array. For instance, sensing circuitry that can include a sense amplifier and a compute component can be coupled to a memory array. A controller can be coupled to the sensing circuitry and can be configured to cause one or more bit strings that are formatted according to a universal number format or a posit format to be transferred from the memory array to the sensing circuitry. The sensing circuitry can perform an arithmetic operation, a logical operation, or both using the one or more bit strings.

An execution unit comprising a processing pipeline configured to perform calculations to evaluate a plurality of mathematical functions. The processing pipeline comprises a plurality of stages through which each calculation for evaluating a mathematical function progresses to an end result. Each of a plurality of processing circuits in the pipeline is configured to perform an operation on input values during at least one stage of the plurality of stages. The plurality of processing circuits include multiplier circuits. A first multiplier circuit and a second multiplier circuit are configured to operate in parallel, such that at the same stage in the processing pipeline, the first multiplier circuit and the second multiplier circuit perform their processing. A third multiplier circuit is arranged in series with the first multiplier circuit and the second multiplier circuit and processes outputs from the first multiplier circuit and the second multiplier circuit.

A method is provided that includes performing, by a processor in response to a vector sort instruction, sorting of values stored in lanes of the vector to generate a sorted vector, wherein the values in a first portion of the lanes are sorted in a first order indicated by the vector sort instruction and the values in a second portion of the lanes are sorted in a second order indicated by the vector sort instruction; and storing the sorted vector in a storage location.

Embodiments detailed herein relate to systems and methods to load a tile register pair. In one example, a processor includes: decode circuitry to decode a load matrix pair instruction having fields for an opcode and source and destination identifiers to identify source and destination matrices, respectively, each matrix having a PAIR parameter equal to TRUE; and execution circuitry to execute the decoded load matrix pair instruction to load every element of left and right tiles of the identified destination matrix from corresponding element positions of left and right tiles of the identified source matrix, respectively, wherein the executing operates on one row of the identified destination matrix at a time, starting with the first row.