A method includes executing software code comprising a plurality of execute packets; responsive to an execute packet of the software code being executed by a data processor core, advancing a value of a test counter register; and responsive to the value of the test counter register being equal to a terminal value, triggering an event to be handled by the software code.

A system and an accelerator circuit includes an internal memory to store data received a memory associated with a processor and a filter circuit block comprising a plurality of circuit stripes, each circuit stripe including a filter processor, a plurality of filter circuits, and a slice of the internal memory assigned to the plurality of filter circuits, where the filter processor is to execute a filter instruction to read data values from the internal memory based on a first memory address, for each of the plurality of circuit stripes: load the data values in weight registers and input registers associated with the plurality of filter circuits of the circuit stripe to generate a plurality of filter results, and write a result generated using the plurality of filter circuits in the internal memory at a second memory address.

A system may include multiple memory cells to store logical data, age tracking circuitry to track a time since a previous access of a particular memory cell, and control circuitry to access the memory cell. Such access may include a read operation of the memory cell, a write operation to the memory cell, or both. The control circuitry may determine an electrical parameter of the memory cell based at least in part on the tracked time since the previous access of the memory cell.

A memory system may include multiple memory cells to store logical data and cycle tracking circuitry to track a number of cycles associated the memory cells. The cycles may be representative of one or more past accesses of the memory cells. The memory system may also include control circuitry to access the memory cells. Accessing of the memory cell may include a read operation, a write operation, or both. During the accessing of the memory cell, the control circuitry may determine a voltage parameter of the access based at least in part on the tracked number of cycles.

A microprocessor system includes a processing circuit and a memory operably coupled to the processing circuit and configured to receive input data according to a pack and store operation and output the data according to a load and unpack operation. The processing circuit comprises a hardware extension configured to: configure a variable number of bits per data element during a pack and store operation; store a concatenation of a plurality of data elements with a reduced number of bits; extract a plurality of data elements with a reduced number of bits during a load and unpacking operation; and recreate a plurality of data elements with an increased number of bits per data element representative of the data elements prior to the pack and store operation.

Systems, methods, and apparatuses relating to microarchitectural mechanisms for the prevention of side-channel attacks are disclosed herein. In one embodiment, a processor includes a core having a plurality of physical contexts to execute a plurality of threads, a plurality of structures shared by the plurality of threads, a context mapping structure to map context signatures to respective physical contexts of the plurality of physical contexts, each physical context to identify and differentiate state of the plurality of structures, and a context manager circuit to, when one or more of a plurality of fields that comprise a context signature is changed, search the context mapping structure for a match to another context signature, and when the match is found, a physical context associated with the match is set as an active physical context for the core.

Data processing apparatus comprises a processing element configured to access an architectural register representing a given system register; mapping circuitry to map the architectural register representing the given system register to a physical register selected from a set of physical registers; a register bank having a set of two or more respective banked versions of the given system register, in which a respective one of the banked versions of the system register is associated with each of a plurality of current operating states of the processing element; in which, when the processing element changes operating state from a first operating state associated with a first one of the banked versions of the system register to a second operating state associated with a second, different, one of the banked versions of the system register, the processing element is configured to store the current contents of the architectural register representing the given system register to the first one of the banked versions of the system register and to copy the contents of the second one of the banked versions of the system register to the architectural register representing the given system register.

A system includes a processor and an accelerator circuit including an input circuit block comprising an input processor to perform first tasks of the neural network application, a filter circuit block comprising a filter processor to perform second tasks of the neural network application, and a plurality of general-purpose filters communicatively coupled to the input circuit block, the filter circuit block, where the input circuit block and the filter circuit block form stages of an execution pipeline, a producer stage is to supply data values to a consumer stage, and operation of the consumer stage is on hold until a start flag stored in a first general-purpose register of the plurality of general-purpose registers to be set by the producer stage.

A system and method include an accelerator circuit comprising an input circuit block, a filter circuit block, a post-processing circuit block, and an output circuit block and a processor to initialize the accelerator circuit, determining tasks of a neural network application to be performed by at least one of the input circuit block, the filter circuit block, the post-processing circuit block, or the output circuit block, assign each of the tasks to a corresponding one of the input circuit block, the filter circuit block, the post-processing circuit block, or the output circuit block, instruct the accelerator circuit to perform the tasks, and execute the neural network application based on results received from the accelerator circuit completing performance of the tasks.

Performing shadow stack functionality for a thread in an audit mode includes initiating execution of a thread at the processor. Execution of the thread includes initiating execution of executable code of an application binary as part of the thread and enabling shadow stack functionality for the thread in an audit mode. Based at least on the execution of the thread in the audit mode, at least a portion of the shadow stack is enabled to be a circular stack. In response to determining that usage of the shadow stack has reached the defined threshold, one or more currently used entries of the shadow stack are overwritten, preventing the shadow stack from overflowing.