EE385B March 30 M. Flynn, Reconfigurable Architectures [Paper: 1970 Bob Rosen and M. Flynn, Prospects of Reconfigurable Architectures] Obvious applications of PLD/PIC*s: - logic emulation (e.g. for design validation) - support functions (e.g. OS related) - security related functions - diagnostics - embedded functions - ...... However, *user reconfigurability* was not successful so far; by and large because of the lack of a suitable computational model! Time element: If Time to Market is essential, PLD/PIC*s are interesting; however, once a function is found with its marketplace a custom design is in order (e.g. arithmetic functions). *Non-VanNeuman-Machines have been either Misunderstood Future Architectures or Turkeys over the last 30 years!* [?MF] Discussion, Questions: - PLD/PIC*s represent a *Mass of Niche Markets* - Arithmetic functions: suppose we run out of resources and need another arithmetic function unit, does it make sense to use a *Reconfigurable* ALU in addition to conventional hw? [?AZ] - Was *user reconfigurability* so far not successful because of lack of compiler support? - Can meaningful experiments be devised to demonstrate the usefulness of user reconfigurability? - Is user reconfigurability just a *dangerous way to surf the evolving technology wave* (time element)?