Michael Schellenberger Costa

Implement `cuda::mr::cuda_async_memory_resource`

12

This implements two facilities for allocating stream order device accessible memory. The first commit adds `cuda_memory_pool`, a wrapper around `::cudaMemPool_t` that takes care of initialization of the buffer and its...

[FEA]: provide a type erased owning resource wrapper

16

### Is this a duplicate? - [X] I confirmed there appear to be no [duplicate issues](https://github.com/NVIDIA/cccl/issues) for this request and that I agree to the [Code of Conduct](CODE_OF_CONDUCT.md) ### Area...

We realized that we do not have tests for thrust::optional in the repo. We should add them

We are currently guarding the deduction guides by standard year but we should just use the feature flag Question: doesn't this fail for compilers with partial C++17 support and `-std=c++1z`?...

The documentation is lying to me. It says the alias is to `vsmem_helper_t`, but it aliases to `vsmem_helper_fallback_policy_t`. This may be a copy&paste error from above? _Originally posted by @bernhardmgruber...

This is a bug in as we might encounter nested namespaces cuda that will conflict. This should be fully qualified `::cuda::std::` https://github.com/NVIDIA/cccl/blob/b3ef1e6a87e0eee7c382f5e6e474e9cfa1c44c36/cub/cub/agent/agent_three_way_partition.cuh#L96

### Is this a duplicate? - [x] I confirmed there appear to be no [duplicate issues](https://github.com/NVIDIA/cccl/issues) for this request and that I agree to the [Code of Conduct](CODE_OF_CONDUCT.md) ### Area...

[FEA]: consider to use `nv_exec_check_disable` for most of libcu++ algorithms / functors

2

### Is this a duplicate? - [X] I confirmed there appear to be no [duplicate issues](https://github.com/NVIDIA/cccl/issues) for this request and that I agree to the [Code of Conduct](CODE_OF_CONDUCT.md) ### Area...

[BUG]: Investigate whether the thrust deprecated compiler warnings are still in effect

4

### Is this a duplicate? - [X] I confirmed there appear to be no [duplicate issues](https://github.com/NVIDIA/cccl/issues) for this bug and that I agree to the [Code of Conduct](CODE_OF_CONDUCT.md) ### Type...

In order to move our `` plans forward we want to implement a memory resource that utilizes stream ordered allocations through `cudaMallocAsync` and `cudaFreeAsync` to allocate memory. That memory_resource should...