Daniel Jünger

@kkraus14 I can move this up on my task list and hammer out a new draft PR tomorrow so we can get started on discussing the last few design questions....

Superseeded by #573

Thanks for catching this! This would be nice to have for the sake of API consistency. I don't think this would spawn any unforeseen synchronization problems as long as the...

@allisonvacanti suggested to utilize versioned [inline namespaces](https://www.foonathan.net/2018/11/inline-namespaces/#content) to solve this problem: ```cpp namespace cuco { inline namespace CUCO_VERSION { void foo(); } /* inline namespace CUCO_VERSION */ } /* namespace...

Initial prototype: https://godbolt.org/z/Kj763Teon This approach still needs some refinement but enables us to use fully variadic ctors if we opt for strong types for all parameters. We could also use...

New version: tested on gcc, nvcc 11.0-11.7 and has better diagnostics: https://godbolt.org/z/5Ghac54E6

As @jrhemstad mentioned earlier, setting up a CI for this would not be fun.

Hi Adam, Thanks for reaching out. Unfortunately, this project is no longer actively maintained. Most concepts introduced by WarpCore have been merged into the OSS library [NVIDIA/cuCollections](https://github.com/NVIDIA/cuCollections), which is actively...

You can specify additional target architectures, e.g., `compute_80` and `compute_86` for Ampere. However, the default setting is optimized for a DGX-1 (Volta) system - hence the default `compute_70` target. More...

The cuco fix was placed ToT. Unfortunately this branch includes some breaking changes for rapids (which have already been addressed for the 24.08 release). I can provide a bugfix branch...