build(cudf): Bring in ucxx when ucx is present and update all RAPIDS dependencies to latest

CMake/resolve_dependency_modules/cudf.cmake

@@ -17,52 +17,71 @@ include_guard(GLOBAL)
 # 3.30.4 is the minimum version required by cudf
 cmake_minimum_required(VERSION 3.30.4)
 
-# rapids_cmake commit d79e071 from 2026-05-01
+# rapids_cmake commit d69f757 from 2026-05-14
 set(VELOX_rapids_cmake_VERSION 26.06)
-set(VELOX_rapids_cmake_COMMIT d79e071f805e709771b80008d50a8b3a5bed93ca)
+set(VELOX_rapids_cmake_COMMIT d69f7578b64e3ea86d881c2676d49b1c34605c09)
 set(
   VELOX_rapids_cmake_BUILD_SHA256_CHECKSUM
-  d0f9eea4feaef1cc325e86eac787052ec951659fdcf21abdfb06efc337a63179
+  e05431e6e56949b24f68e54276b308df625e999a46f3a6ff699fd862d780dd86
 )
 set(
   VELOX_rapids_cmake_SOURCE_URL
   "https://github.com/rapidsai/rapids-cmake/archive/${VELOX_rapids_cmake_COMMIT}.tar.gz"
 )
 velox_resolve_dependency_url(rapids_cmake)
 
-# rmm commit 2357ddd from 2026-05-04
+# rmm commit 0b710a9 from 2026-05-29
 set(VELOX_rmm_VERSION 26.06)
-set(VELOX_rmm_COMMIT 2357ddddcff042ba378e8de6f89e4a995a23b2db)
+set(VELOX_rmm_COMMIT 0b710a93571115d6c3711932c4f9bcf51c23c82c)
 set(
   VELOX_rmm_BUILD_SHA256_CHECKSUM
-  61492c2da88e7f6a6a4edc7101cce4698c156704d135db0b80119f4b9a2c575c
+  4468c6cea82eae2f6ba18b187d365d07e8227874880fbad365e4537ae325f5ac
 )
 set(VELOX_rmm_SOURCE_URL "https://github.com/rapidsai/rmm/archive/${VELOX_rmm_COMMIT}.tar.gz")
 velox_resolve_dependency_url(rmm)
 
-# kvikio commit 247b64e from 2026-05-02
+# kvikio commit d9717c5 from 2026-05-29
 set(VELOX_kvikio_VERSION 26.06)
-set(VELOX_kvikio_COMMIT 247b64e97ecb7cb9ccb06ab123aea87ac571c5b4)
+set(VELOX_kvikio_COMMIT d9717c56c8ad1fce9f57242e488191d38016b142)
 set(
   VELOX_kvikio_BUILD_SHA256_CHECKSUM
-  6419490de95e412cefdbafffc73fac6b2162bc2200076611883fe41637028198
+  306d4213f56cae320286ead7b50a6004c3139d2948822faf548daee529e62025
 )
 set(
   VELOX_kvikio_SOURCE_URL
   "https://github.com/rapidsai/kvikio/archive/${VELOX_kvikio_COMMIT}.tar.gz"
 )
 velox_resolve_dependency_url(kvikio)
 
-# cudf commit d09d10d from 2026-05-04
+# cudf commit 5ea7e1f from 2026-05-29
 set(VELOX_cudf_VERSION 26.06 CACHE STRING "cudf version")
-set(VELOX_cudf_COMMIT d09d10d14d3ed932b8de93638809101af5c7fec3)
+set(VELOX_cudf_COMMIT 5ea7e1f25e6621b4f7256bc88c8bf1de62f2a292)
 set(
   VELOX_cudf_BUILD_SHA256_CHECKSUM
-  5042ec46beb8260eb60d13b9cd44f26357b9756628f7d58659c77e88c67e15d5
+  ec7726869c5cd43d793bef1e0b652d149870a4783fe601797ee8e8302d81cae1
 )
 set(VELOX_cudf_SOURCE_URL "https://github.com/rapidsai/cudf/archive/${VELOX_cudf_COMMIT}.tar.gz")
 velox_resolve_dependency_url(cudf)
 
+# Probe for a system UCX install. The variables are used only to gate ucxx
+# fetching below; nothing in Velox links against UCX directly yet.
+find_library(UCX_LIBRARY NAMES ucp)
+find_path(UCX_INCLUDE_DIR NAMES ucp/api/ucp.h)
+if(UCX_LIBRARY AND UCX_INCLUDE_DIR)
+  message(STATUS "Found UCX: ${UCX_LIBRARY} (headers: ${UCX_INCLUDE_DIR}) -- ucxx will be fetched")
+  # ucxx commit 2e37c84 from 2026-05-29 (release v0.49.00)
+  set(VELOX_ucxx_VERSION 0.50)
+  set(VELOX_ucxx_COMMIT 2e37c8463544064e680e51820c47bfec69f55b69)
+  set(
+    VELOX_ucxx_BUILD_SHA256_CHECKSUM
+    b0172dd8278f95c85f68ba9d35a01445965f33edb6ea3e344c7cb7e0cb8ce3e2
+  )
+  set(VELOX_ucxx_SOURCE_URL "https://github.com/rapidsai/ucxx/archive/${VELOX_ucxx_COMMIT}.tar.gz")
+  velox_resolve_dependency_url(ucxx)
+else()
+  message(STATUS "UCX not found -- ucxx will not be fetched")
+endif()
+
 # Use block so we don't leak variables
 block(SCOPE_FOR VARIABLES)
   # Setup libcudf build to not have testing components
@@ -104,8 +123,23 @@ block(SCOPE_FOR VARIABLES)
     UPDATE_DISCONNECTED 1
   )
 
+  if(UCX_LIBRARY AND UCX_INCLUDE_DIR)
+    FetchContent_Declare(
+      ucxx
+      URL ${VELOX_ucxx_SOURCE_URL}
+      URL_HASH ${VELOX_ucxx_BUILD_SHA256_CHECKSUM}
+      SOURCE_SUBDIR
+      cpp
+      UPDATE_DISCONNECTED 1
+    )
+  endif()
+
   FetchContent_MakeAvailable(cudf)
 
+  if(UCX_LIBRARY AND UCX_INCLUDE_DIR)
+    FetchContent_MakeAvailable(ucxx)
+  endif()
+
   # cudf sets all warnings as errors, and therefore fails to compile with velox
   # expanded set of warnings. We selectively disable problematic warnings just for
   # cudf

scripts/setup-centos-adapters.sh

@@ -25,13 +25,13 @@
 #     Use "n" to never wipe directories.
 # * VELOX_CUDA_VERSION="12.9": Which version of CUDA to install, will pick up
 #   CUDA_VERSION from the env
-# * VELOX_UCX_VERSION="1.19.0": Which version of ucx to install, will pick up
+# * VELOX_UCX_VERSION="1.20.1": Which version of ucx to install, will pick up
 #   UCX_VERSION from the env
 
 set -efx -o pipefail
 
 VELOX_CUDA_VERSION=${CUDA_VERSION:-"12.9"}
-VELOX_UCX_VERSION=${UCX_VERSION:-"1.19.0"}
+VELOX_UCX_VERSION=${UCX_VERSION:-"1.20.1"}
 SCRIPT_DIR=$(dirname "${BASH_SOURCE[0]}")
 source "$SCRIPT_DIR"/setup-centos9.sh
 

scripts/update-cudf-deps.sh

@@ -127,15 +127,38 @@ get_version() {
   fi
 }
 
+get_ucxx_version() {
+  # UCXX uses its own library version instead of the cuDF release version.
+  local ref=$1
+  local response
+  response=$(fetch_github_api "repos/rapidsai/ucxx/contents/VERSION?ref=${ref}") || return 1
+  echo "$response" | jq -r '.content' | base64 -d | grep -oP '^[0-9]+\.[0-9]+' || {
+    echo "Error: Failed to parse VERSION file from ucxx:$ref" >&2
+    return 1
+  }
+}
+
+get_ucxx_branch() {
+  local branch=$1 version=$2
+  if [[ $branch =~ ^release/ ]]; then
+    local ucxx_version
+    # This maps from cuDF release branches (YY.MM) to the corresponding UCXX version.
+    ucxx_version=$(curl -sL --max-time 30 "https://version.gpuci.io/rapids/${version}") || return 1
+    echo "release/${ucxx_version}"
+  else
+    echo "$branch"
+  fi
+}
+
 update_dependency() {
-  local var=$1 commit=$2 date=$3 checksum=$4
+  local var=$1 commit=$2 date=$3 checksum=$4 version=$5
   sed -i "s/# ${var} commit [a-f0-9]* from [0-9-]*/# ${var} commit ${commit:0:7} from ${date}/" "$CMAKE_FILE"
   sed -i "s/set(VELOX_${var}_COMMIT [a-f0-9]*)/set(VELOX_${var}_COMMIT ${commit})/" "$CMAKE_FILE"
 
   if [[ $var == "cudf" ]]; then
-    sed -i "s/set(VELOX_${var}_VERSION [0-9.]* CACHE/set(VELOX_${var}_VERSION ${VERSION} CACHE/" "$CMAKE_FILE"
+    sed -i "s/set(VELOX_${var}_VERSION [0-9.]* CACHE/set(VELOX_${var}_VERSION ${version} CACHE/" "$CMAKE_FILE"
   else
-    sed -i "s/set(VELOX_${var}_VERSION [0-9.]*)/set(VELOX_${var}_VERSION ${VERSION})/" "$CMAKE_FILE"
+    sed -i "s/set(VELOX_${var}_VERSION [0-9.]*)/set(VELOX_${var}_VERSION ${version})/" "$CMAKE_FILE"
   fi
 
   awk -v var="VELOX_${var}_BUILD_SHA256_CHECKSUM" -v sum="$checksum" '
@@ -160,7 +183,7 @@ if [[ $MODE == "--pr" ]]; then
   echo "  SHA256: $CHECKSUM"
   echo
 
-  update_dependency "cudf" "$SHA" "$DATE" "$CHECKSUM"
+  update_dependency "cudf" "$SHA" "$DATE" "$CHECKSUM" "$VERSION"
   echo "Done! Updated cudf to PR #${ARG}: ${SHA:0:7} ($DATE)"
 
 elif [[ $MODE == "--commit" ]]; then
@@ -175,14 +198,15 @@ elif [[ $MODE == "--commit" ]]; then
   echo "  Version: $VERSION"
   echo
 
-  declare -A COMMITS DATES CHECKSUMS
+  declare -A COMMITS DATES CHECKSUMS VERSIONS
   COMMITS[cudf]=$SHA
   DATES[cudf]=$DATE
+  VERSIONS[cudf]=$VERSION
 
   echo "Finding compatible dependency versions (main branch commits before $TIMESTAMP)..."
   echo
 
-  for dep in rapids_cmake rmm kvikio; do
+  for dep in rapids_cmake rmm kvikio ucxx; do
     repo=${dep//_/-}
     echo "Fetching $repo..."
     read -r commit date < <(get_commit_before_date "$repo" "$TIMESTAMP") || exit 1
@@ -194,6 +218,10 @@ elif [[ $MODE == "--commit" ]]; then
     COMMITS[$dep]=$commit
     DATES[$dep]=$date
     CHECKSUMS[$dep]=$checksum
+    VERSIONS[$dep]=$VERSION
+    if [[ $dep == "ucxx" ]]; then
+      VERSIONS[$dep]=$(get_ucxx_version "$commit") || exit 1
+    fi
     echo
   done
 
@@ -203,27 +231,32 @@ elif [[ $MODE == "--commit" ]]; then
   echo
 
   echo "Updating $CMAKE_FILE..."
-  for dep in rapids_cmake rmm kvikio cudf; do
-    update_dependency "$dep" "${COMMITS[$dep]}" "${DATES[$dep]}" "${CHECKSUMS[$dep]}"
+  for dep in rapids_cmake rmm kvikio ucxx cudf; do
+    update_dependency "$dep" "${COMMITS[$dep]}" "${DATES[$dep]}" "${CHECKSUMS[$dep]}" "${VERSIONS[$dep]}"
   done
 
   echo "Done! Updated dependencies:"
-  for dep in rapids_cmake rmm kvikio cudf; do
+  for dep in rapids_cmake rmm kvikio ucxx cudf; do
     echo "  $dep: ${COMMITS[$dep]:0:7} (${DATES[$dep]})"
   done
 
 elif [[ $MODE == "--branch" ]]; then
   VERSION=$(get_version "$ARG") || exit 1
+  UCXX_BRANCH=$(get_ucxx_branch "$ARG" "$VERSION") || exit 1
   echo "Updating cuDF dependencies from branch $ARG (version $VERSION)"
   echo
 
-  declare -A COMMITS DATES CHECKSUMS
+  declare -A COMMITS DATES CHECKSUMS VERSIONS
 
-  for dep in rapids_cmake rmm kvikio cudf; do
+  for dep in rapids_cmake rmm kvikio ucxx cudf; do
     repo=${dep//_/-}
     echo "Fetching $repo..."
 
-    read -r commit date < <(get_commit_info "$repo" "$ARG") || exit 1
+    if [[ $dep == "ucxx" ]]; then
+      read -r commit date < <(get_commit_info "$repo" "$UCXX_BRANCH") || exit 1
+    else
+      read -r commit date < <(get_commit_info "$repo" "$ARG") || exit 1
+    fi
     echo "  Commit: ${commit:0:7} from $date"
     echo "  Computing SHA256..."
     checksum=$(get_sha256 "$repo" "$commit") || exit 1
@@ -232,16 +265,20 @@ elif [[ $MODE == "--branch" ]]; then
     COMMITS[$dep]=$commit
     DATES[$dep]=$date
     CHECKSUMS[$dep]=$checksum
+    VERSIONS[$dep]=$VERSION
+    if [[ $dep == "ucxx" ]]; then
+      VERSIONS[$dep]=$(get_ucxx_version "$commit") || exit 1
+    fi
     echo
   done
 
   echo "Updating $CMAKE_FILE..."
-  for dep in rapids_cmake rmm kvikio cudf; do
-    update_dependency "$dep" "${COMMITS[$dep]}" "${DATES[$dep]}" "${CHECKSUMS[$dep]}"
+  for dep in rapids_cmake rmm kvikio ucxx cudf; do
+    update_dependency "$dep" "${COMMITS[$dep]}" "${DATES[$dep]}" "${CHECKSUMS[$dep]}" "${VERSIONS[$dep]}"
   done
 
   echo "Done! Updated dependencies:"
-  for dep in rapids_cmake rmm kvikio cudf; do
+  for dep in rapids_cmake rmm kvikio ucxx cudf; do
     echo "  $dep: ${COMMITS[$dep]:0:7} (${DATES[$dep]})"
   done
 else