Hip SYCL is new AdaptiveCPP and CUDA 12.6 with clang 18

[stefano2734]

Why change established names here? But it is changed.

https://github.com/AdaptiveCpp/AdaptiveCpp

Perhaps a new release of ndzip possible for more performance with some solved things in clang, sycl or new adaptiveCpp, cuda?

CUDA 12.6 with clang 18 support https://docs.nvidia.com/cuda/cuda-toolkit-release-notes/index.html https://docs.nvidia.com/cuda/pdf/CUDA_Toolkit_Release_Notes.pdf

[fknorr]

Have you had any issues building with a recent AdaptiveCpp / CUDA / Clang? AdaptiveCpp still exports the hipSYCL name if I'm not mistaken.

[stefano2734]

Link of hip SYCL goes to AdaptiveCpp, so readme and in documentation hip SYCL should be extended with framework AdaptiveCPP.

SYCL is the branding name by Khronos and AdaptiveCpp is a unnecessary complication of information. Example: https://www.phoronix.com/news/Open3D-v0.19-Released

[stefano2734]

SYCL is now in Revision 9 of Release 2020 with SYCL 1.2.1. https://www.khronos.org/sycl/

So perhaps some improvements are possible with new features and less bugs.

And AdaptiveCpp makes some progress to all platforms. https://github.com/AdaptiveCpp/AdaptiveCpp/releases/tag/v24.10.0

And experimental with cuda, perhaps more possible with newer hardware and Cuda 12.x.

[stefano2734]

With actual hipCL alias adaptiveCPP 25.02 also running on NVIDIA and Intel.

https://github.com/AdaptiveCpp/AdaptiveCpp/releases/tag/v25.02.0