Design space exploration of network processor architectures

摘要: Design Space Exploration of Network Processor Architectures possible alternatives from the design space, which are evaluated using our calculus, and the feedback guides further exploration.To speed up the exploration, unlike previous approaches, we use several linear approximations in the real-time calculus so that the different system properties can be quickly estimated. We also show how different resources with possibly different scheduling strategies, and communication resources with different arbitration mechanisms, can be combined to construct a scheduling network, which allows us to determine, among other things, the size of shared as well as perresource memory. Our multiobjective design space exploration takes into account the fact that there can be different scenarios in which the processor may be deployed, and this is modeled in the form of different usage scenarios. Lastly, the way we allocate the multiple processing units and the memory units, our optimization strategy also optimizes the load balancing between them. Most of the previous work on modeling, performance evaluation, and design space exploration of network processors (such as [5] and [4]) relied on simulation techniques, where different architectures are simulated and evaluated using benchmark workloads. The work in [5] and [4] addresses issues related to identifying appropriate workloads and modeling frameworks to aid full system analysis and evaluation using simulation. An analytical performance model for network processors was proposed in [21] and [9]. Different network processor architectures can be evaluated on benchmark workloads using this …