Pydgin provides a collection of classes and functions which act as an embedded architectural description language (embedded-ADL) for concisely describing the behavior of instruction set simulators (ISS). An ISS described in Pydgin can be directly executed in a Python interpreter for rapid prototyping and debugging, or alternatively can be used to automatically generate a performant, JIT-optimizing C executable more suitable for application development.
Automatic generation of JIT-enabled ISS from Pydgin is enabled by the RPython Translation Toolchain, an open-source tool used by developers of the PyPy JIT-optimizing Python interpreter.
An ISS described in Pydgin implements an interpretive simulator which can be directly executed in a Python interpreter for rapid prototyping and debugging. However, Pydgin ISS can also be automatically translated into a C executable implementing a JIT-enabled interpretive simulator, providing a high-performance implementation suitable for application development. Generated Pydgin executables provide significant performance benefits in two ways. First, the compiled C implementation enables much more efficient execution of instruction-by-instruction interpretive simulation than the original Python implementation. Second, the generated executable provides a trace-JIT to dynamically compile frequently interpreted hot loops into optimized assembly.
