Task allocation system and method for operation of graphic processing unit

The present invention provides a task allocation system and method for an operation of a graphic processing unit (GPU), which adjust the optimized number of cores for the graphic processing unit on the basis of a memory response time received from the graphic processing unit in allocating a task to the graphic processing unit and processing the allocated task by the graphic processing unit. The present invention allows the optimized number of cores to work in the graphic processing unit, thus reducing the task processing delay due to a memory bottleneck phenomenon while maintaining the task processing speed of the graphic processing unit.