Intel® Academic Program for oneAPI

“We found that porting optimized Hi-Chi code to DPC++ can be done quite quickly by replacing the way that parallel computing is employed. After appropriate tuning, the DPC++ implementation is only slightly inferior to the optimized C++ code, while it became possible to run it on CPUs and GPUs. It turned out that the performance of the DPC++ code on Intel GPUs met expectations even without additional GPU-specific optimizations that could however lead to additional improvements”

— Dr. Iosif Meyerov, professor at Lobachevsky State University of Nizhny Novgorod

“Porting a large application such as NAMD from CUDA to DPC++ is a daunting task. The Intel DPC++ Compatibility Tool (DPCT) reduces much of the manual work by automatically converting calls to the CUDA API to corresponding DPC++ functionalities. This allows the programmer to focus on ensuring the correctness of the application, with minimal effort in manually writing DPC++ code.”

— Dr. David Hardy, professor at the University of Illinois Urbana-Champaign

“oneAPI and DPC++ are enablers for rapid realization of data parallel algorithms and the toolset is powerful!”

— Dr. Yan Luo, University of Massachusetts Lowell