A grid provides abundant resources to the grid users. Task scheduling is a fundamental issue in grid computing. The objective of task scheduling is to allocate required resources to user request. Grid application that requires fast task execution does not perform well since tasks are assigned according to node availability not according to node computing capability. In this paper, we discussed adaptive resource clustering architecture that virtually grouping same computing capability nodes based on the number and resource requirement of tasks so that task execution becomes faster. In this paper, we evaluate the performance of adaptive clustering with static and without clustering of resources and task are schedule by Max-Min, Min-Min, FCFS heuristics and simulation results shows that our architecture outperforms in makespan and success execution rate of tasks.