Coded caching has been shown to significantly reduce traffic load for wireless content delivery by creating coded multicasting opportunities. However, its order-optimal memory-load tradeoff performance requires the file size to go to infinity or grow exponentially with the number of users. This letter introduces a new delivery scheme for decentralized cache placement in practical settings with finite file packetization. The main idea is to characterize the fitness level of each pair of mergable packets, then choose the packet pair with the minimum misfit value for coded multicasting in a greedy and sequential manner. Numerical results show that the proposed scheme attains the lowest traffic load among all existing schemes at all values of file size and cache size with polynomial complexity.