Clustering is an effective method to increase the available parallelism in VLIW datapaths without incurring severe penalties associated with a large number of register file ports. Efficient utilization of a clustered datapath requires careful binding/assignment of operations to clusters. The article proposes a binding algorithm that effectively explores trade-offs between in-cluster operation serialization and delays associated with data transfers between clusters. Extensive experimental evidence is provided showing that the algorithm generates high quality solutions for representative kernels, with up to 33% improvement over a state-of-the-art binding algorithm.

Abstract-Specialized clustered very large instruction word (VLIW) processors combined with effective compilation techniques enable aggressive exploitation of the high instruction-level parallelism inherent in many embedded media applications, while unlocking a variety of possible performance/cost tradeoffs. In this work, the authors propose a methodology to support early design space exploration of clustered VLIW datapaths, in the context of a specific target application. They argue that, due to the large size and complexity of the design space, the early design space exploration phase should consider only design space parameters that have a first-order impact on two key physical figures of merit: clock rate and power dissipation. These parameters were found to be: maximum cluster capacity, number of clusters, and bus (interconnect) capacity. Experimental validation of their design space exploration algorithm shows that a thorough exploration of the complex design space can be performed very efficiently in this abstract parameterized design space. Moreover, an empirical study carried out on a representative set of computation-intensive benchmarks suggests that "penalties" of clustered versus centralized datapaths are often minimal and that clustering indeed unlocks a variety of valuable design tradeoffs.

No abstract

No abstract

Clustering is an effective method to increase the available parallelism in VLIW datapaths without incurring severe penalties associated with large number of register file ports. Efficient utilization of a clustered datapath requires careful binding of operations to clusters. The paper proposes a binding algorithm that effectively explores tradeoffs between in-cluster operation serialization and delays associated with data transfers between clusters. Extensive experimental evidence is provided showing that the algorithm generates high quality solutions for basic blocks, with up to 29% improvement over a state-of-the-art advanced binding algorithm.