B-ASIC: a framework for simulation and implementation of static DSP algorithms

This talk introduces the open source B-ASIC toolbox. B-ASIC is a Python toolbox to assist in implementation of static DSP algorithms. Starting from a block diagram/signal flow graph, that the user can define using arbitrary operations, simulations can be done to verify the correctness. With computational properties of the involved operations, the user can schedule the execution of the operations, both manually and automatically, and finally map computations and storage to the corresponding resources.

B-ASIC can be seen either as a tool for keeping track of manual implementation steps or as a high-level synthesis tool with more control and the ability to manually control and interrupt the stages.

There is still work to do to get a complete path from algorithm to HDL. Still, although the current version require a bit of manual RTL-writing, it is still useful to define and understand architectures. This also makes it suitable for teaching mapping algorithms to hardware ("VLSI signal processing").