Techniques and systems for prescheduling an alternative CPU as soon as a need for a task is detected by a primary CPU are disclosed. A process includes detecting, by a producer thread running on a first CPU, an external interrupt, acquiring, by the producer thread, a spinlock, and in response to acquiring the spinlock, sending, by the producer thread, an event to a consumer thread that is to run on a second CPU. Upon receiving the event by the consumer thread, the consumer thread acquires the spinlock, which “stalls” the consumer thread in a loop until the spinlock is released by the producer thread. While the consumer thread is “waking up” in response to receiving the event from the producer thread, the producer thread creates a task, publishes the task, and releases the spinlock, thereby causing the consumer thread to exit the loop, retrieve the task, and execute the task.

A method for minimizing lock contention among threads in a multithreaded system is disclosed. The method includes the steps of: (a) a processor causing a control thread, if information on a task is acquired by the control thread, to acquire a lock to thereby put the information on a task into a specific task queue which satisfies a certain condition among multiple task queues; and (b) the processor causing a specified worker thread corresponding to the specific task queue among multiple worker threads, if the lock held by the control thread is released, to acquire a lock to thereby get a task stored in the specific task queue.

Described are techniques for avoiding deadlock. Each consumer is assigned a priority level denoting a number of resource allocation requests that have been currently granted to the consumer. The priority level is incremented for each resource allocation request granted to the consumer. A set of common resource pool(s) and a reserved resource pool are provided. A wait list of pending resource requests made by the consumers is maintained. If no allocated resources of the set of common pool(s) have been released for at least a threshold amount of time and there is at least one request on the wait list, first processing is performed which includes selecting, in accordance with one or more criteria, a first pending resource request from the wait list that was made by a first consumer, and assigning the first consumer exclusive use of the reserved pool for a lifetime of the first consumer.

The subject technology performs a locking operation on a first set of keys by a first statement of a first transaction. The subject technology determines that a conflict occurred between the first statement and a second transaction. The subject technology determines that the second transaction has yet to complete after a predetermined period of time. The subject technology performs a deadlock detection process where the subject technology stores a key and value in a table indicating the first transaction and the second transaction, detects, based at least in part on a graph traversal of the table starting from the first transaction, a cycle between the first transaction and the second transaction, and determines that the first transaction is a youngest transaction in the detected cycle. The subject technology ceases execution of the first transaction in response to the first transaction being a youngest transaction in a detected cycle.

A method for improving the performance of applications executed within asynchronous processor architectures. In an embodiment, a method for improving execution time of compiled synchronized source code on an asynchronous processor architecture includes receiving, by a processing system, synchronized source code comprising synchronization instructions to synchronize execution of the synchronized source code on different pipelines of the asynchronous processor architecture. The method also includes analyzing, by the processing system, the synchronized source code to determine whether the synchronized source code includes a broken code condition. The method also includes, after determining, by the processing system, that the synchronized source code does not include a broken code condition, outputting an optimized synchronized source code generated by performing a corrective action on the synchronized source code to correct a synchronization inaccuracy, inconsistency, or inefficiency in the synchronized source code.

A first and a second catalog access statement are obtained. For the first catalog access statement, a first lock schedule which includes a catalog resource being locked and a lock strength is obtained. For the second catalog access statement, a second lock schedule which includes a catalog resource being locked and a lock strength are obtained. A potential deadlock between the first and the second catalog access statement is identified, including by comparing the first lock schedule and the second lock schedule.

Techniques for managing register allocation are provided. The techniques include detecting a first request to allocate first registers for a first wavefront; first determining, based on allocation information, that allocating the first registers to the first wavefront would result in a condition in which a deadlock is possible; in response to the first determining, refraining from allocating the first registers to the first wavefront; detecting a second request to allocate second registers for a second wavefront; second determining, based on the allocation information, that allocating the second registers to the second wavefront would result in a condition in which deadlock is not possible; and in response to the second determining, allocating the second registers to the second wavefront.

A method, system, and apparatus are provided for accelerated atomic resource allocation on a multiprocessor platform. In particular, a resource allocation engine (RAE) performs the following: counting available units for each of the one or more resources; parsing a multi-resource ticket (MRT) for a processor, wherein the parsing identifies one or more requested resource types, each resource type being paired with a requested resource units; comparing the multi-resource ticket to one or more resource queues for the requested resource types, wherein the comparing determines an availability status of at least one the requested resource types; and based on the availability status, calculating whether or not all of the requested resource types can be allocated for the processor, wherein the calculating is completed before allocating a next requested resource for a next processor.

A semiconductor device performs a software lock-step. The semiconductor device includes a first circuit group including a first Intellectual Property (IP) to be operated in a first address space, a first bus, and a first memory, a second circuit group including a second IP to be operated in a second address space, a second bus, and a second memory, a third bus connectable to a third memory, and a transfer control circuit coupled to the first to third buses. when the software lock-step is performed, the second circuit group converts an access address from the second IP to the second memory such that an address assigned to the second memory in the second address space is a same as an address assigned to the first memory in the first address space.

A method for executing multithreaded instructions grouped into blocks. The method includes receiving an incoming instruction sequence using a global front end; grouping the instructions to form instruction blocks, wherein the instructions of the instruction blocks are interleaved with multiple threads; scheduling the instructions of the instruction block to execute in accordance with the multiple threads; and tracking execution of the multiple threads to enforce fairness in an execution pipeline.