Abstract: Systems, apparatuses, and methods for compressing multiple instruction operations together into a single retire queue entry are disclosed. A processor includes at least a scheduler, a retire queue, one or more execution units, and control logic. When the control logic detects a given instruction operation being dispatched by the scheduler to an execution unit, the control logic determines if the given instruction operation meets one or more conditions for being compressed with one or more other instruction operations into a single retire queue entry. If the one or more conditions are met, two or more instruction operations are stored together in a single retire queue entry. By compressing multiple instruction operations together into an individual retire queue entry, the retire queue is able to be used more efficiently, and the processor can speculatively execute more instructions without the retire queue exhausting its supply of available entries.

Abstract: Software instructions are executed on a processor within a computer system to configure a streaming engine with stream parameters to define a multidimensional array. The stream parameters define a size for each dimension of the multidimensional array and a pad value indicator. Data is fetched from a memory coupled to the streaming engine responsive to the stream parameters. A stream of vectors is formed for the multidimensional array responsive to the stream parameters from the data fetched from memory. A padded stream vector is formed that includes a specified pad value without accessing the pad value from system memory.

Abstract: Techniques for operating on an indirect memory access instruction, where the instruction accesses a memory location via at least one indirect address. A pipeline processes the instruction and a memory operation engine generates a first access to the at least one indirect address and a second access to a target address determined by the at least one indirect address. A cache memory used with the pipeline and the memory operation engine caches pointers. In response to a cache hit when executing the indirect memory access instruction, operations dereference a pointer to obtain the at least one indirect address, not set a cache bit, and return data for the instruction without storing the data in the cache memory; and in response to a cache miss, operations set the cache bit, obtain, and store a cache line for a missed pointer, and return data without storing the data in the cache memory.

Abstract: Systems and methods for preserving the quality of service for client applications having workloads for execution by a compute core or a hardware accelerator are described. A method for operating a hardware accelerator configured to process commands submitted by client applications to the hardware accelerator, where a workload can be executed either by a compute core or by the hardware accelerator, is described. The method includes queueing commands for execution of workloads in a first set of command queues and queueing commands for execution of workloads in a second set of command queues. The method includes workload processors executing workloads specified by commands in the first set of command queues and the second set of command queues in an order of execution that is determined based on output of a set of trackers configured to track one or more criteria for a selected set of command queues.

Abstract: The present disclosure describes a system and method for preempting a long-running process with a higher priority process in a machine learning system, such as a hardware accelerator. The machine learning hardware accelerator can be a multi-chip system including semiconductor chips that can be application-specific integrated circuits (ASIC) designed to perform machine learning operations. An ASIC is an integrated circuit (IC) that is customized for a particular use.

Abstract: Methods and systems for transmitting data are presented. Data received from at least one data source is retained in at least one buffer. In one example, initial hierarchical data may be provided from the at least one buffer to a device, followed by additional hierarchical data. In one example, the data is received into the at least one buffer via a multicast connection, and the data is provided to the device via a point-to-point connection.

Abstract: A nested loop controller includes a first register having a first value initialized to an initial first value, a second register having a second value initialized to an initial second value, and a third register configured as a predicate FIFO, initialized to have a third value. The second value is advanced in response to a tick instruction during execution of a loop. In response to the second value reaching a second threshold, the second register is reset to the initial second value. The nested loop controller further includes a comparator coupled to the second register and to the predicate FIFO and configured to provide an outer loop indicator value as input to the predicate FIFO when the second value is equal to the second threshold, and provide an inner loop indicator value as input to the predicate FIFO when the second value is not equal to the second threshold.

Abstract: Techniques are disclosed for communicating between a machine learning accelerator and one or more processing cores. The techniques include obtaining data at the machine learning accelerator via an input/output die; processing the data at the machine learning accelerator to generate machine learning processing results; and exporting the machine learning processing results via the input/output die, wherein the input/output die is coupled to one or more processor chiplets via one or more processor ports, and wherein the input/output die is coupled to the machine learning accelerator via an accelerator port.

Abstract: An audience measurement computing system for monitoring a media presentation device in a monitored environment is described and includes a network interface, at least one processor, and a non-transitory computer-readable medium comprising instructions executable by the processor(s). The computing system is configured to obtain, via a cable connected to an input port of the media presentation device, a voltage signal generated by the media presentation device based on an operational state of the media presentation device; compare voltage indicated by the voltage signal to a threshold; based on the comparing, generate timestamped operational state data comprising a record indicative of when the media presentation device is in an on-state; obtain audience measurement data representing one or more media signals communicated to the media presentation device; and transmit, via the network interface over a network and to a central facility, the timestamped operational state data and the audience measurement data.

Abstract: A method in a reconfigurable computing system includes receiving a user program for execution on a reconfigurable dataflow computing system, comprising a grid of compute units and grid of memory units interconnected with a switching array. The user program includes multiple tensor-based algebraic expressions that are converted to an intermediate representation comprising one or more logical operations executable via dataflow through compute units. These one or more logical operations are preceded by or followed by a buffer, each buffer corresponding to one or more memory units. The method includes determining whether splitting a selected buffer yields a reduced cost and then splitting the selected buffer, in response to the determining step, to produce first and second buffers. Dataflow through memory units corresponding to the first and second buffers is controlled by one or more memory units within the grid of memory units. Buffer splitting optimization reduces memory unit consumption.

Abstract: An audience measurement computing system for monitoring a media presentation device in a monitored environment is described and includes a network interface, at least one processor, and a non-transitory computer-readable medium comprising instructions executable by the processor(s). The computing system is configured to obtain, via a cable connected to an input port of the media presentation device, a voltage signal generated by the media presentation device based on an operational state of the media presentation device; compare voltage indicated by the voltage signal to a threshold; based on the comparing, generate timestamped operational state data comprising a record indicative of when the media presentation device is in an on-state; obtain audience measurement data representing one or more media signals communicated to the media presentation device; and transmit, via the network interface over a network and to a central facility, the timestamped operational state data and the audience measurement data.

Abstract: Aspects relate to lane failure recovery for a data link having multiple lanes labeled in a contiguous sequence. In one aspect, a failure of a failed lane of the data link is detected. Working lanes of the data link are then detected. A set of contiguous working lanes of the data link are selected, and an operational link as including the selected set of contiguous working lanes is defined. A start address of the operational link is identified and stored in a configuration register. Data traffic is transmitted on the operational link.

Abstract: A system for performing energy-efficient computing reduces the amount of data that is transferred between a processor and an external memory device. The processor and the external memory device are equipped with first and second near data processing control units (NCUs), respectively, that coordinate offloading of preselected subprocesses from the processor to a first processing circuit disposed on or near the external memory device. When the processor is performing one of these preselected processes, the first NCU transmits commands and memory addresses to the second NCU. The processing circuit on or near the memory device performs the subprocess or subprocesses and the result is forwarded by the second NCU to the first NCU, which forwards it to the processor to complete the process.

Abstract: In accordance with described techniques for processing-in-memory (PIM) search stop control, a computing system or computing device includes a memory system that includes a stop condition check component, which receives an instruction that includes a programmed check value. The stop condition check component compares the programmed check value to outputs of a PIM component, and the stop condition check component initiates a stop instruction to stop the PIM component from processing subsequent computations based on an output of the PIM component matching the programmed check value.

Abstract: Embodiments detailed herein relate to matrix operations. In particular, matrix (tile) multiply accumulate and negated matrix (tile) multiply accumulate are discussed. For example, in some embodiments decode circuitry to decode an instruction having fields for an opcode, an identifier for a first source matrix operand, an identifier of a second source matrix operand, and an identifier for a source/destination matrix operand; and execution circuitry to execute the decoded instruction to multiply the identified first source matrix operand by the identified second source matrix operand, add a result of the multiplication to the identified source/destination matrix operand, and store a result of the addition in the identified source/destination matrix operand and zero unconfigured columns of identified source/destination matrix operand are detailed.

Abstract: A method for scheduling input/output (I/O) commands is described. The method includes receiving, by an I/O scheduler, an I/O command from an application; generating, by the I/O controller, an I/O resource requirement based on the I/O command; determining, by a traffic controller, that an amount of available resources satisfies a criteria based on the I/O resource requirement; and sending, by the traffic controller, the I/O command to a queue in response to the criteria being satisfied.

Abstract: A parallel processing unit comprises a plurality of processors each being coupled to a memory access hardware circuitry. Each memory access hardware circuitry is configured to receive, from the coupled processor, a memory access request specifying a coordinate of a multidimensional data structure, wherein the memory access hardware circuit is one of a plurality of memory access circuitry each coupled to a respective one of the processors; and, in response to the memory access request, translate the coordinate of the multidimensional data structure into plural memory addresses for the multidimensional data structure and using the plural memory addresses, asynchronously transfer at least a portion of the multidimensional data structure for processing by at least the coupled processor. The memory locations may be in the shared memory of the coupled processor and/or an external memory.

Abstract: A system on chip, semiconductor device, and/or method are provided that include a plurality of masters, an interface, and a semaphore unit. The interface interfaces the plurality of masters with a slave device. The semaphore unit detects requests of the plurality of masters, controlling the salve device, about an access to the interface and assigns a semaphore about each of the plurality of masters by a specific operation unit according to the detection result.

Abstract: Methods and apparatus for improved data movement operations through interconnect fabric. In one embodiment, Non-Transparent Bridge (NTB) technology used to perform data movement operations between a host and multiple peer devices using a DMA (direct memory access) engine and at least one descriptor ring having enhanced descriptor entries. In one implementation, descriptor ring entries include source and destination address information, address translation information, and fabric partition information. In one implementation, a DMA engine is configured directly access host memory and generate data packets using the descriptor entry information. In one embodiment, the descriptor ring is a virtual descriptor ring located on DMA hardware, host memory, or elsewhere in the NT fabric address space, and may be accessed by user processes.

Abstract: A PLC according to an embodiment of the present invention is disclosed. The PLC according to an embodiment of the present invention comprises: a master unit; and a plurality of interface units which access the master unit via a system bus. The master unit includes: a control unit for controlling input/output and processing commands; and a plurality of first RJ45 terminals for input/output of data, wherein a maximum of eight signals are arranged via each pin of each of the first RJ45 terminals. Each of the interface units includes a connection means for connecting a signal between a second RJ45 terminal for connecting a signal line from each of the first RJ45 terminals and an external device, wherein the interface units are configured to be expandable by the number of the first RJ45 terminals.