To fully exploit multicore processors, processes are expected to provide thread-level parallelism with the aim of improving the performance of the application and achieving efficient hardware utilization for parallel workloads. One of the important issues of OSes is how to distribute processes among cores to achieve some performance goals, fairness in this context where fairness is an essential criterion of any operating system scheduler. This distribution is performed by transferring tasks (called load balancing) from the heavily loaded cores to the lightly loaded cores. Current task scheduler of the Linux kernel, Completely Fair Scheduler (CFS), uses weight-based load balancing mechanism to evenly distribute tasks’ weights among all cores. Unfortunately, CFS scheduler is non-scalable and inaccurate for multicore processors. In this paper, we propose a thread-level scheduler which presents a load balancing technique designed specifically for the purpose of overcoming the shortage in achieving multicore scheduling and fair-share together for parallel processes running on multicore systems. Instead of balancing run queue length, our algorithm balances the time of threads which are running on “light” and “heavy” cores. To demonstrate its effectiveness, we have implemented our algorithm in Linux kernel and demonstrate its efficacy through extensive experiments. Our evaluation results show that our scheduler achieves better fairness and high performance. We conclude from our results that our scheduler is practical and desirable for general purpose multicore processors.