A couple of high level impressions.
Here my notes from the “Managing Project Health” BoF at DAC 2009.
- After sitting through the presentations back to back I saw the outline of a four layer model for design management
- data stability – essentially what companies like Cliosoft or open source tools like CVS offer
- provisioning – what Flex-LM for license enablement and LSF for machine allocation offer
- flow definition and management – where Lynx and Runtime Design focus
- project level status and semantics – Achilles, SiBridge, and SatinIP are wrestling with different aspects of this.
- The Mentor Xtreme environment addressed all four layers at once and enabled real time shared write access. I think this is a very under-appreciated application.
- Runtime presented an anonymized customer flow with 26,000 jobs in a complex mesh, and showed a table of applications that included a 500,000 job library verification flow. There was a question from the audience as to why any of this was necessary with “make -j” available. The answer was in an environment with homogeneous resources and an homogeneous job mix (e.g. a typical verification environment running a lot of random seed simulations) “make -j” was serviceable. If either the resources or job requirements were complex it would not give as good an outcome.
- The Runtime Workload Analyzer enabled what if analysis, an existing job mix could be replayed against a computing environment with a different set of computing resources or software licenses.
- The Achilles heat map that showed a year’s worth of project status on one dashboard generated a lot of discussion.
- Cliosoft and Runtime are more than 10 years old, Xtreme is more than 6, Achilles is four years old (although it was their first year at DAC). None of these problems are new and there is an interaction between project organization structure, multi-firm collaboration