Intrepydd: performance, productivity, and portability for data science application kernels

This is a presentation of our talk at Onward 2020. We introduce the Intrepydd programming system, which enables data scientists to write application kernels with high performance, productivity, and portability on current and future hardware. To deliver high performance, the Intrepydd toolchain uses ahead-of-time (AOT) compilation and high-level compiler optimizations. Intrepydd achieves portability by its ability to compile kernels for execution on different hardware platforms, and for invocation from Python or C++ main programs. An empirical evaluation shows significant performance improvements relative to Python, and the suitability of Intrepydd for mapping on to post-Moore accelerators and architectures with relative ease. We believe that Intrepydd represents a new direction of ``Discipline-Aware Languages'' (DiALs), which brings us closer to the holy grail of obtaining productivity and portability and more generality than current domain-specific languages and libraries.

Presentation Videos