Open
Description
Description
C would be equivalent to the current FAST_RUN. In the future we may want to change FAST_RUN to be the current NUMBA
Python would be new, something with a python VM but fast_run optimizations and no cxx optimizations. May consider if we want to use the JIT linker machinery to define python code that avoids the input/output storage hoops that exist right now in Op.perform. Ideally this would be as fast as running sequential numpy code, with the only overhead being a function call for each Op.