Merge branch 'branch-24.04' into fix-cagra-prune-illegal-memory-access

.pre-commit-config.yaml

@@ -81,12 +81,6 @@ repos:
                 verbose: true
                 require_serial: true
                 exclude: .*/thirdparty/.*
-              - id: copyright-check
-                name: copyright-check
-                entry: python ./ci/checks/copyright.py --git-modified-only --update-current-year
-                language: python
-                pass_filenames: false
-                additional_dependencies: [gitpython]
               - id: include-check
                 name: include-check
                 entry: python ./cpp/scripts/include_checker.py cpp/bench cpp/include cpp/test
@@ -109,6 +103,23 @@ repos:
         rev: v4.5.0
         hooks:
               - id: check-json
+      - repo: https://github.com/rapidsai/pre-commit-hooks
+        rev: v0.0.1
+        hooks:
+          - id: verify-copyright
+            files: |
+              (?x)
+                  [.](cmake|cpp|cu|cuh|h|hpp|sh|pxd|py|pyx)$|
+                  CMakeLists[.]txt$|
+                  CMakeLists_standalone[.]txt$|
+                  meta[.]yaml$|
+                  setup[.]cfg$
+            exclude: |
+              (?x)
+                  cpp/include/raft/neighbors/detail/faiss_select/|
+                  cpp/include/raft/thirdparty/|
+                  docs/source/sphinxext/github_link[.]py|
+                  cpp/cmake/modules/FindAVX[.]cmake
 
 default_language_version:
       python: python3

build.sh

@@ -305,7 +305,7 @@ if hasArg --allgpuarch; then
     BUILD_ALL_GPU_ARCH=1
 fi
 
-if hasArg --compile-lib || (( ${NUMARGS} == 0 )); then
+if hasArg --compile-lib || hasArg pylibraft || (( ${NUMARGS} == 0 )); then
     COMPILE_LIBRARY=ON
     CMAKE_TARGET="${CMAKE_TARGET};raft_lib"
 fi
@@ -405,7 +405,7 @@ fi
 
 ################################################################################
 # Configure for building all C++ targets
-if (( ${NUMARGS} == 0 )) || hasArg libraft || hasArg docs || hasArg tests || hasArg bench-prims || hasArg bench-ann; then
+if (( ${NUMARGS} == 0 )) || hasArg libraft || hasArg docs || hasArg tests || hasArg bench-prims || hasArg bench-ann || ((${COMPILE_LIBRARY} == ON )); then
     if (( ${BUILD_ALL_GPU_ARCH} == 0 )); then
         RAFT_CMAKE_CUDA_ARCHITECTURES="NATIVE"
         echo "Building for the architecture of the GPU in the system..."

conda/recipes/libraft/build_libraft_template.sh

@@ -1,5 +1,5 @@
 #!/usr/bin/env bash
-# Copyright (c) 2022-2023, NVIDIA CORPORATION.
+# Copyright (c) 2022-2024, NVIDIA CORPORATION.
 
 # Just building template so we verify it uses libraft.so and fail if it doesn't build
-./build.sh template
+./build.sh template --no-nvtx

cpp/CMakeLists.txt

@@ -256,7 +256,7 @@ endif()
 
 if(RAFT_NVTX)
   # This enables NVTX within the project with no option to disable it downstream.
-  target_link_libraries(raft INTERFACE CUDA::nvToolsExt)
+  target_link_libraries(raft INTERFACE CUDA::nvtx3)
   target_compile_definitions(raft INTERFACE NVTX_ENABLED)
 else()
   # Allow enable NVTX downstream if not set here. This creates a new option at build/install time,
@@ -324,6 +324,7 @@ if(RAFT_COMPILE_LIBRARY)
     src/distance/detail/pairwise_matrix/dispatch_russel_rao_float_float_float_int.cu
     src/distance/distance.cu
     src/distance/fused_l2_nn.cu
+    src/distance/fused_distance_nn.cu
     src/linalg/detail/coalesced_reduction.cu
     src/matrix/detail/select_k_double_int64_t.cu
     src/matrix/detail/select_k_double_uint32_t.cu
@@ -422,6 +423,7 @@ if(RAFT_COMPILE_LIBRARY)
     src/raft_runtime/cluster/update_centroids.cuh
     src/raft_runtime/cluster/update_centroids_double.cu
     src/raft_runtime/cluster/update_centroids_float.cu
+    src/raft_runtime/distance/fused_distance_min_arg.cu
     src/raft_runtime/distance/fused_l2_min_arg.cu
     src/raft_runtime/distance/pairwise_distance.cu
     src/raft_runtime/matrix/select_k_float_int64_t.cu