It is important to write portable programs while concealing various latencies that would be problematic in getting the optimal performance out of a parallel processing environment. Fine Grain Parallelism is one of the methods for implementing this. However, there is a problem in achieving Fine Grain Parallelism using a conventional OS. This is because, as the granularity becomes finer, the frequency of context switching, scheduling, and the like increases, leading to an increase in the overhead due to those factors, and there would be a risk of causing severe performance degradation. Therefore, we propose a multiprocessor architecture using a Scheduling Support Hardware (SSH), which aims to achieve high-speed context switching and scheduling while utilizing Fine Grain Parallelism, by having the hardware support the thread scheduling and the allocation/release of the CPU resources, which are part of the function of the OS. In this paper, we designed a multiprocessor system with SSH using the Verilog-HDL and showed the effectiveness of our approach in an evaluation by a computer simulation. In the evaluation, we found that the scheduling time could be reduced by using the SSH, and thus it became possible to use a parallelism with a finer granularity. © 2006 Wiley Periodicals, Inc. Syst Comp Jpn, 37(2): 79–95, 2006; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10224