Intel Wants To Contribute Parallel STL Support To libstdc++ / libc++

The last major item for GCC's libstdc++ standard library for C++17 support is supporting the technical specification around parallelism and Intel is hoping to land their implementation of it for both libstdc++ and libc++.

Alexey Kukanov of Intel's expert compiler team posted a "request for comments" on Wednesday about contributing the company's implementation of the C++17 parallel algorithms for libstdc++. A similar query is also being sent to the LLVM folks working on libc++.

Intel's code is currently available via GitHub for supporting the C++17 standard execution policies and the experimental unsequenced policy for SIMD execution.

The developers are building upon Intel's Threading Building Blocks (TBB) for the thread-level parallelism but is abstracted with an internal API that could allow libstdc++/libc++ developers to also support other threading APIs. None of their current code is Intel CPU specific.

They are still tidying up this Parallel STL code and ensuring it follows all portions of the specification but for now are seeking initial comments via the libstdc++ mailing list. So far there is a fair amount of enthusiasm about this contribution for getting this important technical specification supported by the key Linux C++ standard libraries. The TS is documented as N4659.