- METHOD FOR SIMPLIFIED TASK-BASED RUNTIME FOR EFFICIENT PARALLEL COMPUTING

摘要

Embodiments include computing devices, systems, and methods for implementing scheduling and execution of lightweight kernels as simple tasks directly by threads without setting up a task structure. The computing device may determine whether the task pointer in the task queue is a simple task pointer to the lightweight kernel. The computing device may schedule a first simple task for a lightweight kernel for execution by a thread. The computing device may retrieve a kernel pointer to the lightweight kernel from an entry in the simple task table. An entry in the simple task table may be associated with a simple task pointer. The computing device may also execute the lightweight kernel directly as a simple task.;