A pipeline bubbles reduction technique for the Monsoon dataflow architecture

This paper proposes two types of new auxiliary matching stores (AMS) that can be applied to dynamic dataflow machines with a frame-structured memory. The first type of the auxiliary matching store achieves high matching rate by combining the LRU replacement algorithm and a simple match structure together. The second one achieves high match rate by controlling strictly the number of active processes with an explicit hardware. Owing to some special characteristics of dynamic dataflow computing, this temporary storage must be designed based on the following principles: (a) The number of the active processes must be limited. (b) The related instructions must be arranged nearly. (c) The replacement of the match slot must be deterministic.<>