Compiler Support

As of today (July 2019) only two compilers/STL implementation support parallel algorithms: it’s Visual Studio (since 2017 15.7) and GCC (since 9.1).

Visual Studio implements par_unseq as par, so you shouldn’t expect any difference between code runs.

GCC implementation uses modified Intel PSTL and relies on OpenMP 4.0 and Intel TBB 2018. You need to install and link with -ltbb if you want to work with parallel algorithms.

For example:

g++-9.1 -std=c++17 -Wall -O2 sample.cpp -ltbb

For Building GCC 9.1 and Intel TBB you can check this guide @Solarian Programmer: C++17 STL Parallel Algorithms - with GCC 9.1 and Intel TBB on Linux and macOS.

Summary:

Feature	GCC	Clang	MSVC
Parallel Algorithms	9.1[^pargcc]	in progress	VS 2017 15.7[^parstlmsvc]

[^pargcc]: See in the article: GCC 9.1 Released and GCC 9 Release Series — Changes, New Features, and Fixes) [^parstlmsvc]: See Announcing: MSVC Conforms to the C++ Standard | Visual C++ Team Blog

There are also several other implementations out there:

• Codeplay - SYCL Parallel STL
• STE||AR Group - HPX
• Intel - Parallel STL - based on OpenMP 4.0 and Intel® TBB.
• Parallel STL - early Microsoft implementation for the Technical Specification.
• n3554 - proposal implementation (started by Nvidia)
• Thibaut Lutz Implementation @Github - early implementation

---

In the next chapter, we will look at other changes in the library.