During the past decade or so software developers have been bound to CUDA to write highly parallel software for GPUs. The challenge with this approach is that CUDA is a proprietary programming interface and can only be used with Nvidia GPUs. The landscape for processors is changing rapidly with vendors, including Intel, and new specialized processors such as those harnessing the RISC-V ISA bringing increasing choice.
This session will show how easy it is to migrate your code from CUDA and port it to SYCL enabling your organization to be free to make a vendor neutral choice while still supporting your existing GPU hardware. Through practical examples we will explain the steps to semi-automatically port your code from CUDA to SYCL along with some tips to achieve the best performance on your target processor.