| 摘要 |
A method and system is described for synchronizing execution by a processing element of threads within a process. Before execution of a thread commences, a determination is made as to whether all of the required resources for execution of the thread are available in a cache local to the processing element. If the resources are not available, then the resources are fetched from main storage and stored in one or more local caches before execution begins. If the resources are available, then execution of the thread may begin. During execution of the thread and, in particular, an instruction within the thread, the instruction may require data in order to successfully complete its execution. When this occurs, a determination is made as to whether the necessary data is available. If the data is available, the result of the instruction execution is stored and execution of the thread continues. However, if the data is unavailable, then the thread is deferred until the data becomes available and a new thread is processed. When deferring a thread, the thread is placed in the memory location which is to receive the required data. Once the data is available, the thread is removed from the data location and placed on a queue for execution and the data is stored in the location. <IMAGE> |
