Remove cutlass and cuco dependencies (rapidsai#2916)

As the `distance` APIs are now moved to cuVS, we can remove the cutlass and cuco dependencies and avoid shipping the headers.

Closes rapidsai#2914 

Reduces CUDA 13 libraft wheel size from 18.9M to 14.5M.

Authors:
  - Divye Gala (https://github.com/divyegala)
  - Bradley Dice (https://github.com/bdice)

Approvers:
  - Bradley Dice (https://github.com/bdice)
  - Dante Gama Dessavre (https://github.com/dantegd)
  - Corey J. Nolet (https://github.com/cjnolet)

URL: rapidsai#2916

Author: divyegala

.pre-commit-config.yaml

@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2022-2025, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2022-2026, NVIDIA CORPORATION.
 # SPDX-License-Identifier: Apache-2.0
 
 repos:
@@ -83,7 +83,6 @@ repos:
                 exclude: |
                   (?x)
                     ^CHANGELOG[.]md$|
-                    ^cpp/cmake/patches/cutlass/build-export[.]patch$
       - repo: https://github.com/pre-commit/pre-commit-hooks
         rev: v5.0.0
         hooks:

conda/recipes/libraft/recipe.yaml

@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2025, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2025-2026, NVIDIA CORPORATION.
 # SPDX-License-Identifier: Apache-2.0
 schema_version: 1
 
@@ -90,9 +90,8 @@ outputs:
     build:
       script:
         content: |
-          # Unspecified brings in cuco and cutlass and friends
+          # Unspecified brings in all components
           cmake --install cpp/build --component Unspecified
-          cmake --install cpp/build --component cuco
           cmake --install cpp/build --component raft
           cmake --install cpp/build --component distributed
       string: cuda${{ cuda_major }}_${{ date_string }}_${{ head_rev }}

cpp/.clang-format

@@ -85,7 +85,7 @@ IncludeCategories:
     Priority:        5
   - Regex:           '^<(thrust|cub|cuda)/' # CCCL includes
     Priority:        6
-  - Regex:           '^<(cooperative_groups|cuco|cuda|device_types|driver_types|math_constants|nvtx3)' # CUDA includes
+  - Regex:           '^<(cooperative_groups|cuda|device_types|driver_types|math_constants|nvtx3)' # CUDA includes
     Priority:        6
   - Regex:           '^<.*\..*' # other system includes (e.g. with a '.')
     Priority:        7

cpp/CMakeLists.txt

@@ -1,6 +1,6 @@
 # =============================================================================
 # cmake-format: off
-# SPDX-FileCopyrightText: Copyright (c) 2020-2025, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2020-2026, NVIDIA CORPORATION.
 # SPDX-License-Identifier: Apache-2.0
 # cmake-format: on
 cmake_minimum_required(VERSION 3.30.4 FATAL_ERROR)
@@ -164,13 +164,9 @@ include(${rapids-cmake-dir}/cpm/rapids_logger.cmake)
 rapids_cpm_rapids_logger(BUILD_EXPORT_SET raft-exports INSTALL_EXPORT_SET raft-exports)
 create_logger_macros(RAFT "raft::default_logger()" include/raft/core)
 
-# CCCL before rmm/cuco so we get the right version of CCCL
+# CCCL before rmm so we get the right version of CCCL
 include(cmake/thirdparty/get_cccl.cmake)
 include(cmake/thirdparty/get_rmm.cmake)
-include(cmake/thirdparty/get_cutlass.cmake)
-
-include(${rapids-cmake-dir}/cpm/cuco.cmake)
-rapids_cpm_cuco(BUILD_EXPORT_SET raft-exports INSTALL_EXPORT_SET raft-exports)
 
 if(BUILD_TESTS)
   include(${rapids-cmake-dir}/cpm/gtest.cmake)
@@ -193,10 +189,7 @@ target_include_directories(
 )
 
 # Keep RAFT as lightweight as possible. Only CUDA libs and rmm should be used in global target.
-target_link_libraries(
-  raft INTERFACE rapids_logger::rapids_logger rmm::rmm cuco::cuco nvidia::cutlass::cutlass
-                 CCCL::CCCL
-)
+target_link_libraries(raft INTERFACE rapids_logger::rapids_logger rmm::rmm CCCL::CCCL)
 
 target_compile_features(raft INTERFACE cxx_std_17 $<BUILD_INTERFACE:cuda_std_17>)
 target_compile_options(

cpp/cmake/patches/cutlass/Support-both-CUDA-12-and-13-cccl-header-locations.patch

@@ -86,8 +86,6 @@ In addition to the libraries included with cudatoolkit 12.2+, there are some oth
 #### Required
 - [RMM](https://github.com/rapidsai/rmm) corresponding to RAFT version.
 - [Thrust](https://github.com/NVIDIA/thrust) v1.17 / [CUB](https://github.com/NVIDIA/cub)
-- [cuCollections](https://github.com/NVIDIA/cuCollections) - Used in `raft::sparse::distance` API.
-- [CUTLASS](https://github.com/NVIDIA/cutlass)  v2.9.1 - Used in `raft::distance` API.
 
 #### Optional
 - [NCCL](https://github.com/NVIDIA/nccl) - Used in `raft::comms` API and needed to build `raft-dask`.
@@ -110,7 +108,7 @@ The recommended way to build and install RAFT from source is to use the `build.s
 
 ### Header-only C++
 
-`build.sh` uses [rapids-cmake](https://github.com/rapidsai/rapids-cmake), which will automatically download any dependencies which are not already installed. It's important to note that while all the headers will be installed and available, some parts of the RAFT API depend on libraries like CUTLASS, which will need to be explicitly enabled in `build.sh`.
+`build.sh` uses [rapids-cmake](https://github.com/rapidsai/rapids-cmake), which will automatically download any dependencies which are not already installed.
 
 The following example will download the needed dependencies and install the RAFT headers into `$INSTALL_PREFIX/include/raft`.
 ```bash
@@ -288,8 +286,8 @@ PROPERTIES CXX_STANDARD                        20
 
 The `raft::raft` CMake target is made available when including RAFT into your CMake project but additional CMake targets can be made available by adding to the `COMPONENTS` option in CMake's `find_package(raft)` (refer to [CMake docs](https://cmake.org/cmake/help/latest/command/find_package.html#basic-signature) to learn more). The components should be separated by spaces. The `raft::raft` target will always be available. Note that the `distributed` component also exports additional dependencies.
 
-| Component   | Target              | Description                                              | Base Dependencies                      |
-|-------------|---------------------|----------------------------------------------------------|----------------------------------------|
-| n/a         | `raft::raft`        | Full RAFT header library                                 | CUDA toolkit, RMM, NVTX, CCCL, CUTLASS |
-| compiled    | `raft::compiled`    | Pre-compiled template instantiations and runtime library | raft::raft                             |
-| distributed | `raft::distributed` | Dependencies for `raft::comms` APIs                      | raft::raft, UCX, NCCL
+| Component   | Target              | Description                                              | Base Dependencies             |
+|-------------|---------------------|----------------------------------------------------------|-------------------------------|
+| n/a         | `raft::raft`        | Full RAFT header library                                 | CUDA toolkit, RMM, NVTX, CCCL |
+| compiled    | `raft::compiled`    | Pre-compiled template instantiations and runtime library | raft::raft                    |
+| distributed | `raft::distributed` | Dependencies for `raft::comms` APIs                      | raft::raft, UCX, NCCL         |

cpp/cmake/patches/cutlass_override.json

@@ -1,6 +1,6 @@
 # =============================================================================
 # cmake-format: off
-# SPDX-FileCopyrightText: Copyright (c) 2022-2025, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2022-2026, NVIDIA CORPORATION.
 # SPDX-License-Identifier: Apache-2.0
 # cmake-format: on
 # =============================================================================
@@ -9,8 +9,7 @@ cmake_minimum_required(VERSION 3.30.4 FATAL_ERROR)
 
 include(../../cmake/rapids_config.cmake)
 
-# We always need CUDA for pylibraft because the raft dependency brings in a header-only cuco
-# dependency that enables CUDA unconditionally.
+# We always need CUDA for pylibraft because the raft dependency enables CUDA unconditionally.
 include(rapids-cuda)
 rapids_cuda_init_architectures(pylibraft)
 
@@ -20,7 +19,7 @@ project(
   LANGUAGES CXX CUDA
 )
 
-# an installed version of raft contains the other necessary targets (like CCCL and cuco)
+# an installed version of raft contains the other necessary targets (like CCCL)
 find_package(raft "${RAPIDS_VERSION}" REQUIRED COMPONENTS raft compiled)
 
 include(rapids-cython-core)