high performance computing on graphics processing units: hgpu.org

OpenCL SYCL is a new heterogeneous and parallel programming framework created by the Khronos Group that tries to bring OpenCL programming into C++. In particular, it enables C++ developers to create OpenCL kernels, using all the popular C++ features, such as classes, inheritance and templates. What is more, it dramatically reduces programming effort and complexity, by letting the runtime system handle system resources and data transfers. In this project we investigate Syclone, a prototype SYCL implementation that we acquired from Codeplay Ltd. To do that we compare SYCL in terms of both programmability and performance with another two established parallel programming models, OpenCL and OpenMP. In particular, we have chosen six scientific applications and ported them in all three models. Then we run benchmarking experiments to assess their performance on an Intel Xeon Phi based platform. The results have shown that SYCL does great from a programmability perspective. However, its performance still does not match that of OpenCL or OpenMP.