Michael Schellenberger Costa

> It may have broken one test: > > ``` > third_party/gpus/cccl/libcudacxx/test/libcudacxx/std/utilities/meta/meta.unary/meta.unary.prop/is_swappable.pass.cpp:99:23: error: static assertion failed due to requirement '!cuda::std::is_swappable::value': > 99 | static_assert(!cuda::std::is_swappable::value, ""); > | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > 1...

> Looks like CL is green again. Would you be able to update your backport branch, so I can test the current changes with v2.3.2? Done

Oh, I believe you are seeing a mismatch between installed driver and the toolkit you are using. Looking at your `nvidia-smi` output you are running with > CUDA Version: 12.2...

@lw also to be sure, can you verify that you also properly added the `nv` subfolder from the libcudacxx folder to the include path? we did add changes to the...

> If you take the RMM approach of avoiding mixed ownership semantics, then your current `cuda_async_memory_resource`, since it is such a thin wrapper and only decides whether to use the...

To reiterate the recent discussion we had about async memory operations. 1. A memory resource is a low level tool. We are fine with it having sharp edges and users...

@harrism Would love to get your input here

Regarding the mentioned two use cases A way to store the memory resource used per device (shamelessly stolen from rmm) ```cpp using namespace cuda::mr; inline auto& get_map() { static std::map...

In general any RAII type memory functionality kind of wants to actually store the used memory resource and not rely on `get_current_device_resource` because that could have actually changed since it...

Yeah this is explicitly about owning a resource