How to Port Code from CUDA* to SYCL*

This session helps you understand how to port your CUDA* code to SYCL* and continue to target NVIDIA* GPUs while retaining a good level of performance. Using N-body simulation project code written in CUDA, the presenters show how the code is automatically translated to SYCL and then compiled using the Intel® oneAPI DPC++ Compiler. Get performance tips and tricks that ensure the best performance from your SYCL code on NVIDIA GPUs.


Joe Todd is a senior software engineer at Codeplay* with a decade of experience developing parallel software. Joe's career began with a PhD in glaciology at the University of Cambridge where he implemented an ice-fracture extension to the parallel finite element model Elmer FEM. He later spent several years as a postdoctoral researcher at the University of St. Andrews and the University of Edinburgh. Recently, driven by a desire to focus on software engineering, Joe left academia to join Codeplay. He works on the SYCL and CUDA back ends with a continuing focus on particle simulators.