1
کامپیوتر و شبکه::
رابطهٔ نگاشت
However, in previous ILP-based mapping approaches [5], [12] aiming at high performance, the mapping problem is usually formulated by two key binary variables: the mapping relation of a task and each processor and the mapping relation of each pair of tasks and each processor.
Moreover, the ILP-based methods mentioned above all utilize the task-processor mapping relation and task-pair- processor mapping relation as key variables, while none of them consider the communication-processors mapping rela- tion variable as an optimization.
Based on the system model and the application model, task mapping can be modeled as a mapping relation M between a graph set tt and a processor set S denoted by M : tt S, where each task from the targeted application is mapped onto its appropriate processor determining the computation workload on each processor, and precedent dependencies between tasks on different processors are mapped as inter- processor communications.
As the three factors greatly affect the final performance, we build them into the objective function and set constraints for mapping relations and communication pipeline allocation.
In our model, there are two key binary variables MTi,k and MCi,jk representing the task-processor mapping relation and communication-processors mapping relation, respectively.
واژگان شبکه مترجمین ایران