Quickly Migrate CUDA* Code to Multiplatform SYCL* Code

Escape vendor lock-in by migrating from CUDA* code to multiplatform C++ with SYCL* code using the Intel® DPC++ Compatibility Tool. This tool transforms 90% to 95% of CUDA code, delivering SYCL source code tailored for optimized performance on hybrid architectures.

The session starts by showing the migration of a simple CUDA sample with a single CUDA source file. It then looks at migrating a complex CUDA project with multiple source files using CMake* files. When this migration completes, learn how to perform any necessary manual migration steps; an analysis of the code suggests further optimizations for running effectively on multiple vendor architectures.

Topics include:

• An overview of standards-based C++ with SYCL, and a discussion of its origins and benefits to developers.
• Demos with real-world examples of the migration concepts, process, and expectations.
• A discussion of common tasks required to complete the migration, including scenarios where manual porting is required.
• An exploration of post-migration performance tuning that may be needed because of differences between CUDA and C++ with SYCL.

Download the stand-alone version of the Intel DPC++ Compatibility Tool or as part of the Intel® oneAPI Base Toolkit. This toolkit provides a core set of tools and libraries for developing high-performance, data-centric applications across diverse architectures.

Find sample applications, resources, and tools to support your migration from CUDA to C++ with SYCL.