Partitioning a lenient parallel language into sequential threads

Multithreading as attractive in a large-scale parallel system since it allows split-phase memory operations and fast context switching between computations without blocking the processor. Performance of multithreaded archifectures depends significantly on the quality of multithreaded codes. In this paper, we describe the enhanced thread formation scheme to produce efficient sequential threads from programs written, in a len.zent parallel language Id-[I]. This scheme features graph partitaoning based 011 only long latency instructions, combination. of multiple switches and merges introducing a generalized switch and merge, thread merging, and redundant arc eliminataon using thread precedence relation. The simulataon results show that our scheme reduces control and branch instructions effectively.