Intel has been working on its OneAPI project for quite some time. The company has now shared more details of the software project — including the launch of a new programming language called “Data Parallel C++ (DPC++).”
The OneAPI initiative is Intel’s effort to deliver a unified programming model that will simplify application development across different computing architectures.
At Intel’s Software Technology Day in London, the company announced the development of the new programming language/dialect as a part of the OneAPI initiative.
DPC++ is being described as an open, cross-industry programming language based on C++ which will serve as an “alternative to single architecture proprietary language.”
While the company hasn’t revealed any other significant details, it confirmed that the SYCL single-source C++ programming standard from The Khronos Group will be used as the basis for DPC++.
Intel first mentioned the OneAPI last December for optimizing code across CPUs / GPUs / FPGAs to ensure “no transistor (is) left behind.”
It sounds a lot like the Heterogeneous System Architecture (HSA) and is said to be their response to NVIDIA’s CUDA ecosystem.
The OneAPI software itself will support both direct programming and API programming. DPC++ falls under the direct programming wing.
For API-based programming, OneAPI will arrive with powerful libraries where library functions are “custom-coded for each target architecture.”
For more details on Data Parallel C++ programming language, we will have to wait for the big reveal of OneAPU which is expected in Q4 2019 when the project enters the beta phase.