The increasing task computation complexity and limited battery has become a serious concern for smartphones. To reduce the task computation delay and save the smartphone battery usage, there have been many efforts to offload the tasks from the mobile device to the remote cloud with a much higher computation ability. However, offloading tasks to cloud will cause extra transmission delays and energy consumption. In reality, timely task execution is very important because the task's utility decays with time. Therefore, there exists a delay-energy trade-off. This paper addresses the aforementioned challenge. The smartphone should take advantage of the cloud in high computation speed so that the smartphone can achieve the maximum utility with limited battery. We get a 2-approximation schedule on expectation by using a LP rounding algorithm. The real trace experiments show the effectiveness of the proposed algorithms.