From data scientists working in Python to those toiling in assembly language, there’s a need for an open, high-performing alternative to Nvidia’s CUDA, says Intel’s Joe Curley.