Speaker
Jim Pivarski
(Princeton University)
Description
Getting data out of one language/framework and into another, mostly Python <--> C++ and Java <--> C++ or Python.
Advanced Numpy tricks for writing into arrays in C++ libraries, into other processes via shared memory, allocating Numpy (and using Python) in NUMA environments, particularly KNL and GPU.
Java to C++/CPython hooks: JNI, JNA, off-heap memory.
On the fly compilation, which often factors into the above.
Will assume familiarity with C/C++, Python, and Java. Since it's an overview of many similar topics, rather than a deep dive into any one of them, students will benefit if they only have familiarity and interest in two of the three languages.
