CODEPLAY released "ComputeCpp" the first GPU accelerated implementation of SYCL, the high-level parallel standard created by Khronos.
The compiler is still in beta and only released for linux, but it makes possible to generate host and device code from singe source C++11 utilizing massively parallel and hierarchically parallel constructs.
The OpenCL 1.2 based technology runs on AMD and Intel devices at the moment, and provides acceleration support for the Eigen linear algebra, the VisionCpp machine vision and the C++17 parallel STL libraries.
The compiler will be available for windows soon, and will be available for commercial and non-commercial users. As a next step, it will build upon OpenCL 2.2 and will support the now provisional SYCL 2.2 specification in an upcoming version.
The compiler is available after registration from the webpage of CODEPLAY.
Accelerated projects...