I. Introduction
Graphics processing units (GPUs) have been present in the computing industry for over 40 years. Their special design allows us to perform many operations simultaneously. However, the resources on the GPU are limited and some operations take much more time than the others. In order to manage the execution of the warps, a scheduler should keep list of warps that can be issued at any time and issue them such that it maximizes the resources utilization and performance of GPU.