In this session we will migrate a slightly more complex CUDA source to SYCL source using the SYCLomatic tool.

We will migrate the Jacobi Iterative algorithm from Nvidia CUDA samples to SYCL using SYCLomatic tool and then compile and run the migrated C++ SYCL code on Intel GPU.

In this session we will understand how CUDA features like Local Memory, Cooperative groups, warp primitives and atomic operations are migrated to SYCL, we will inspect the CUDA and SYCL source and understand how migration was accomplished using SYCLomatic tool. We will also try to manually optimize the migrated SYCL code for performance using SYCL features.