Summary: This paper considers preemptive scheduling on \(m\) parallel machines with learning effect in cloud manufacturing. Each machine has a unit processing cost. The objective is to select a number of machines from the given \(m\) machines such that the makespan is minimized under the constraint that the total cost is not greater than the given cost budget. The learning effect of the machine cost function is considered. A uniform optimal preemptive algorithm is provided for two learning effect functions based on exponential function and power function, respectively.