Bradley Dice

## Summary - Update `run_ctests.sh` to first try the installed test location (CI/conda environments) and fall back to the build directory (devcontainer environments) - Enables testing in devcontainers with `test-cuml-cpp`...

## Summary - Enable static linking of libcudart by default (`CUDA_STATIC_RUNTIME=ON`) - Remove `cuda-cudart` from conda recipe run requirements (no longer needed when statically linked) This is part of a...

CCCL 3.2 resource ref compatibility

1

## Description This PR implements a "bridge" for CCCL 3.2 compatibility while retaining support for resources stored as `device_memory_resource*`. This is a different approach than #2162/#2166 and hopefully will be...

Add RMM User Guide

2

## Description Draft only, not ready for review. This is a placeholder. I plan to revise this heavily and cut up the new docs into multiple PRs. ## Checklist -...

[FEA] Support memory resources from CCCL 3.2

2

**Is your feature request related to a problem? Please describe.** Recently, RAPIDS/CCCL nightlies began [failing](https://github.com/NVIDIA/cccl/actions/runs/16794408362) due to a change in CCCL's memory resources. See https://github.com/NVIDIA/cccl/pull/5313 **Describe the solution you'd like**...

Add experimental cuda_async_pinned_memory_resource

1

## Description Contributes to #2054. Adds a new `cuda_async_pinned_memory_resource` that provides stream-ordered pinned (page-locked) host memory allocation using CUDA 13.0's `cudaMemGetDefaultMemPool` API with `cudaMemAllocationTypePinned`. This parallels the `cuda_async_managed_memory_resource` added in...

[FEA] Properly support alignment argument in `do_allocate` / `do_deallocate`

2

The signature of CCCL's `allocate` and `deallocate` methods accept a `size_t alignment`. However, RMM's current design doesn't do anything with this alignment. We need to update `do_allocate` and `do_deallocate` in...

## Motivation In CCCL < 3.2, we can construct [`resource_ref` types](https://github.com/rapidsai/rmm/blob/9650719248a50ea400c2493d867eb2f9c6d88f20/cpp/include/rmm/resource_ref.hpp) from `device_memory_resource*`. However, the ability to construct a `resource_ref` from a `device_memory_resource*` has been removed from CCCL 3.2 as...

Test with latest CCCL

4

This is a test PR to demonstrate an issue in building with the CCCL `main` branch.

Use pytest-xdist to parallelize tests

2

## Description This is an experiment to parallelize RMM tests. Now that the async MR doesn't prime itself, there should be fewer issues with this. I may need to tweak...