Disclosed in some examples, are systems, methods, devices, and machine readable mediums which use improved dynamic programming algorithms to pack conditional branch instructions. Conditional code branches may be modeled as directed acyclic graphs (DAGs) which have a topological ordering. These DAGs may be used to construct a dynamic programming table to find a partial mapping of one path onto the other path using dynamic programming algorithms.

A method for managing memory leaks in a memory device includes grouping, by a garbage collection system, a plurality of similar memory allocations of the memory device into one or more Unique Fixed Identifiers (UFIs); identifying, by the garbage collection system, one of the one or more UFIs having a highest accumulated memory size and adding each of the plurality of memory allocations in the identified one of the one or more UFIs into a Potential Leak Candidate List (PLCL); identifying, by the garbage collection system, the memory leaks in the memory device by identifying unreferenced memory addresses associated with the plurality of memory allocations in the PLCL; and releasing, by the garbage collection system, the identified unreferenced memory addresses associated with the plurality of memory allocations corresponding to the memory leaks into the memory device.

A plurality of connection patterns is determined based on connectivity data collected by a plurality of agents. Each agent of the plurality of agents is installed on a respective compute node of a plurality of compute nodes. The connectivity data collected by each agent of the plurality of agents includes node-local connectivity data indicating node-local connections for the respective compute node on which the agent is installed. The node-local connections include communications with at least one application entity hosted by the respective compute node. A graph representation that is organized with respect to the at least one application entity hosted by each of the plurality of compute nodes is generated based on the plurality of connection patterns.

Embodiments of systems, methods, and apparatuses for heterogeneous computing are described. In some embodiments, a hardware heterogeneous scheduler dispatches instructions for execution on one or more plurality of heterogeneous processing elements, the instructions corresponding to a code fragment to be processed by the one or more of the plurality of heterogeneous processing elements, wherein the instructions are native instructions to at least one of the one or more of the plurality of heterogeneous processing elements.

An apparatus and method for tagged memory management. For example, one embodiment of a processor comprises: execution circuitry to execute instructions and process data, at least one instruction to generate a system memory access request having a first address pointer; and address translation circuitry to determine whether to translate the first address pointer with or without metadata processing, wherein if the first address pointer is to be translated with metadata processing, the address translation circuitry to: perform a lookup in a memory metadata table to identify a memory metadata value, determine a pointer metadata value associated with the first address pointer, and compare the memory metadata value with the pointer metadata value, the comparison to generate a validation of the memory access request or a fault condition, wherein if the comparison results in a validation of the memory access request, then accessing a set of one or more address translation tables to translate the first address pointer to a first physical address and to return the first physical address responsive to the memory access request.

In an embodiment, a local memory dedicated to one or more hardware accelerators in a system may include at least two portions: a volatile portion and a non-volatile portion. Data that is reused from iteration to iteration of the hardware accelerator (e.g. constants, instruction words, etc.) may be stored in the non-volatile portion. Data that varies from iteration to iteration may be stored in the volatile portion. Both the local memory and the hardware accelerators may be powered down between iterations, saving power. The non-volatile portion need only be initialized at a first iteration, allowing the amount of time that the hardware accelerators and the local memory are powered up to be lessened for subsequent iterations since the reused data need not be reloaded in the subsequent iterations.

Disclosed is an instruction for a programmable atomic transaction that is executed as the last instruction and that terminates the executing thread, waits for all outstanding store operations to finish, clears the programmable atomic lock, and sends a completion response back to the issuing process. This guarantees that the programmable atomic lock is cleared when the transaction completes. By coupling thread termination with clearing the lock bit, this guarantees that the thread cannot terminate without clearing the lock.

A streaming engine employed in a digital data processor specifies a fixed read only data stream defined by plural nested loops. An address generator produces addresses of data elements. A steam head register stores data elements next to be supplied to functional units for use as operands. Stream metadata is stored in response to a stream store instruction. Stored stream metadata is restored to the stream engine in response to a stream restore instruction. An interrupt changes an open stream to a frozen state discarding stored stream data. A return from interrupt changes a frozen stream to an active state.

There is provided a method of representing a three dimensional (3D) object using univariate curves, comprising: receiving an initial definition of a 3D object representation, calculate a covering set of univariate curves, the covering set comprising at least one non-planar univariate curve, wherein the covering set of univariate curves represent the volume of the 3D object within a tolerance requirement, and generating a representation of the 3D object based on the set of univariate curves, wherein the set of univariate curves represent the volume of the 3D object.

Providing atomicity for complex operations using near-memory computing is disclosed. In an implementation, a complex atomic operation is decomposed into a set of sequential operations that is stored in a near-memory instruction store. A memory controller receives a request from a host execution engine to issue the complex atomic operation and initiates execution of the stored set of sequential operations on a near-memory compute unit. The complex atomic operation may be a user-defined complex atomic operation.