MPSOC ’08, Live from Maastricht: Real Men (and Women) Simulate!

I’ve just returned from Maastricht, Netherlands where I attended the 2008 edition of the MPSOC (Multi or Many Processor SOC) conference. My next few blog entries summarize some of the most relevant MPSOC presentations. Professor Paolo Ienne of the Ecole Polytechnique Federale de Lausanne (EPFL) Processor Architecture Laboratory gave the second presentation at MPSOC ’08 on the explosive needs of system simulation in the MPSOC era. He started his talk by stating that computer architecture is now dominated by simulation and embedded system design is now similarly simulation-centric. However, system complexity has already become a design bottleneck in many cases and simulation rarely provides real insight into or strong guarantees of system performance. Current simulation techniques can say that a system is working but do not generally help in understanding why a system design isn’t working. “We need analytic techniques,” said Professor Ienne.

Many such techniques have been developed including synchronous data-flow (SDF) graphs, stochastic automata networks (SAN), event adaptation functions, and real-time calculus (RTC). Of these, Ienne advocates a form of real-time calculus called network calculus, which is a mathematical model based on min-plus algebra. Network calculus is based on deterministic queuing theory and provides worst-case bounds on system behavior.

What followed was a moderately advanced course in network theory…still to advanced for me. However, I did take two important ideas away from this talk:

1. Analytical modeling converts today’s iterative system-design process into one that converges linearly and quickly to a design solution.
2. Analytical models are extremely difficult to develop and we need mathematicians to make them available to system designers. (Good news for my colleague Grant Martin, I think, who’s a mathematician.)