Tutorial. The SYCL parallel programming model

Tutorial. The SYCL parallel programming model

Registration for this event.

Notice: Fundamental Tutorials registration is closed now.

Synopsis:

Instructors:

  • Amanda Sabatini Dufek - Lawrence Berkeley National Laboratory.
  • Douglas Adriano Augusto - Oswaldo Cruz Foundation.

Requirements: Attendees are expected to have some familiarity with C++ and basic concepts of parallel computing.

Time: 4 hours

Description:The upcoming Exascale supercomputers vary considerably when it comes to accelerator architectures and that sheds light on the importance of portable many-core parallel frameworks, such as OpenCL, SYCL, Kokkos, and OpenMP, in order to run parallel applications across different devices without requiring code modifications. This tutorial will cover the SYCL programming model, which will be available in these new Exascale supercomputers. SYCL is a single-source, high-level, C++ parallel programming model for heterogeneous computing systems. In addition, SYCL is an open standard managed by the Khronos Group for the development of portable parallel applications that can support a wide range of parallel devices (e.g. CPUs, GPUs, and FPGAs) from multiple vendors (e.g. Intel, NVIDIA, and AMD).

Attendees are expected to have some familiarity with C++ and basic concepts of parallel computing.

References:

Data Parallel C++: Mastering DPC++ for Programming of Heterogeneous Systems using C++ and SYCL (2021), Reinders, J., Ashbaugh, B., Brodman, J., Kinsner, M., Pennycook, J., Tian, X.

SYCL Overview - The Khronos Group Inc, https://www.khronos.org/sycl