L. He and Z. Liu (PRC)
SMT, front-end, branch predictor, fetch policy
Simultaneous Multithreaded (SMT) processors improve the instruction throughput by allowing fetching and run ning instructions from several threads simultaneously at a single cycle. Accurate branch prediction and efficient in struction fetching are crucial to an SMT processor. In this paper, we construct an effective SMT front-end through combining a value based branch predictor and an instruc tion throughput based fetch policy. The implementation of the front-end is easy, and the hardware needed is less than that of the traditional front-end. Execution-driven simula tion results show that our front-end improves the perfor mance of an SMT processor dramatically. The average speedup of instruction throughput is up to 28% over that of the traditional SMT front-end, which is also higher than that of stream buffer based front-end in the literature.
Important Links:
Go Back
