Intel Contributes Its Parallel STL Implementation To LLVM

A month ago there was word that Intel wanted to contribute their Parallel STL implementation for this C++17 functionality to GCC's libstdc++ and LLVM libc++. As a wonderful open-source Christmas present, Intel's Parallel STL implementation saw its initial commit now under the LLVM umbrella.

Parallel STL provides a portable implementation of threaded and vectorized execution of standard C++ algorithms. Initially the Intel code is tied to OpenMP 4.0 and Threaded Building Blocks. C++17 adds various parallel algorithms to the standard template library allowing developers to speed up their code in a C++-defined portable manner for tapping today's systems with increasing core/thread counts. More details on the original Intel PSTL implementation can be found via software.intel.com.

Last week Intel made its initial contribution of the 18,755 lines of code forming this implementation. Moving forward this code will continue to be worked on under LLVM's normal development procedures / code review process.

This code is living at LLVM/pstl.