At present, the cost of High Performance Computing (HPC) system is very high and recently cloud has been popular for its operation on HPC. In this study, we investigated the price efficiency that will benefit the customer in choosing provider's package. We analyzed relation between prices, times and problem sizes (N) or workloads from High Performance Linpack on Amazon EC2 in Compute Optimize package. Our result showed that at N ⩽ 40320 the lowest package able to run the fit problem sizes will obtain the best price worthiness. On the other hand, at N > 40320 using the highest package means the best price worthiness. For the proportion between increasing price and reduced running time, mostly the increasing money by x times for using higher instance cannot reduce execution time by x times.