SYCL 2020 in hipSYCL: DPC++ features on AMD GPUs, NVIDIA GPUs and CPUs

SYCL 2020 in hipSYCL: DPC++ features on AMD GPUs, NVIDIA GPUs and CPUs

HipSYCL is one of the four major SYCL implementations, with a particular focus on and aggregating hardware support for multivendor hardware provided by those toolchains within one single framework. Recently, hipSYCL has also started adopting DPC++/SYCL 2020 features such as unified shared memory, reductions and more in order to increase portability of code using these features, in particular with respect to AMD GPUs. After a brief introduction to hipSYCL, we will explore how hipSYCL fits into the oneAPI ecosystem as well as what developers can expect from hipSYCL in terms of functionality and performance. In particular, we will focus on SYCL 2020 and DPC++ features such as unified shared memory or group algorithms across hardware supported by hipSYCL

Download Presentation Deck

Watch Now

Aksel Alpay
Aksel Alpay
Heidelberg University
×


Register for the oneAPI DevSummit hosted by UXL:

Register Now