Intel oneDPL(oneAPI DPC++ Library) Offloads C++ To SYCL
To speed up SYCL kernels on the newest CPUs, GPUs, and other accelerators, get started with oneDPL and examine oneDPL code examples right now
you may improve application performance by offloading Parallel STL algorithms to several devices that support the SYCL programming framework
As outlined in ISO/IEC 14882:2017 (often referred to as C++17) and C++20, the Parallel API in Intel oneAPI DPC++ Library (oneDPL) implements the C++ standard algorithms with execution rules
Intel oneAPI DPC++/C++ Compiler, as well as threaded and SIMD execution of these algorithms on Intel processors built on top of OpenMP and oneTBB
When used with the Intel oneAPI DPC++/C++ Compiler, oneDPL speeds up SYCL kernels for accelerated parallel programming on a variety of hardware accelerators and architectures
In contrast, you may use oneDPL‘s Device Selection API to dynamically assign available computing resources to your workload in accordance
Using the fsycl-pstl-offload option with the Intel oneAPI DPC++/C++ Compiler, it exploits an experimental oneDPL capability