https://hgpu.org/?p=27976
Porting numerical integration codes from CUDA to oneAPI: a case study