Provide 25.03 version bump (#424)

* Provide 25.03 version bump

* Fix build

Author: hcho3

CMakeLists.txt

@@ -27,14 +27,14 @@ set(TRITON_BACKEND_REPO_TAG "main" CACHE STRING "Tag for triton-inference-server
 
 # Specify *minimum* version for all RAPIDS dependencies
 # Some RAPIDS deps may have later versions
-set(RAPIDS_DEPENDENCIES_VERSION "24.10" CACHE STRING "RAPIDS projects dependencies version")
+set(RAPIDS_DEPENDENCIES_VERSION "25.02" CACHE STRING "RAPIDS projects dependencies version")
 set(RAPIDS_TRITON_REPO_PATH "https://github.com/rapidsai/rapids-triton.git" CACHE STRING "Git repository to pull rapids_triton from")
 set(RAPIDS_TRITON_REPO_TAG "branch-${RAPIDS_DEPENDENCIES_VERSION}" CACHE STRING "Tag for rapidsai/rapids-triton repo")
 
 if(TRITON_FIL_DOCKER_BUILD)
-  project(RAPIDS_TRITON_BACKEND VERSION 25.01.00)
-  set(TRITON_BUILD_CONTAINER "nvcr.io/nvidia/tritonserver:25.01-py3" CACHE STRING "Build image for Dockerized builds")
-  set(TRITON_BUILD_CONTAINER_VERSION "25.01" CACHE STRING "Triton version for Dockerized builds")
+  project(RAPIDS_TRITON_BACKEND VERSION 25.02.00)
+  set(TRITON_BUILD_CONTAINER "nvcr.io/nvidia/tritonserver:25.02-py3" CACHE STRING "Build image for Dockerized builds")
+  set(TRITON_BUILD_CONTAINER_VERSION "25.02" CACHE STRING "Triton version for Dockerized builds")
 
   add_custom_command(
     OUTPUT fil/libtriton_fil.so $<$<BOOL:${TRITON_ENABLE_GPU}>:fil/libcuml++.so>
@@ -83,7 +83,7 @@ else()
   ##############################################################################
   # - Prepare rapids-cmake -----------------------------------------------------
   file(DOWNLOAD
-    https://raw.githubusercontent.com/rapidsai/rapids-cmake/branch-24.10/RAPIDS.cmake
+    https://raw.githubusercontent.com/rapidsai/rapids-cmake/branch-25.02/RAPIDS.cmake
       ${CMAKE_BINARY_DIR}/RAPIDS.cmake)
   include(${CMAKE_BINARY_DIR}/RAPIDS.cmake)
   include(rapids-cmake)
@@ -123,9 +123,9 @@ else()
 
   if(TRITON_ENABLE_GPU)
     rapids_cuda_init_architectures(RAPIDS_TRITON_BACKEND)
-    project(RAPIDS_TRITON_BACKEND VERSION 22.10.00 LANGUAGES CXX CUDA)
+    project(RAPIDS_TRITON_BACKEND VERSION 25.03.00 LANGUAGES CXX CUDA)
   else()
-    project(RAPIDS_TRITON_BACKEND VERSION 22.10.00 LANGUAGES CXX)
+    project(RAPIDS_TRITON_BACKEND VERSION 25.03.00 LANGUAGES CXX)
   endif()
 
   ##############################################################################

cmake/thirdparty/get_cuml.cmake

@@ -55,8 +55,8 @@ endfunction()
 # Change pinned tag here to test a commit in CI
 # To use a different RAFT locally, set the CMake variable
 # CPM_raft_SOURCE=/path/to/local/raft
-find_and_configure_cuml(VERSION    24.12
+find_and_configure_cuml(VERSION    25.02
                         FORK       rapidsai
-                        PINNED_TAG feature/fil-backend-r25.01
+                        PINNED_TAG branch-25.02
                         USE_TREELITE_STATIC ${TRITON_FIL_USE_TREELITE_STATIC}
                         )

conda/environments/triton_benchmark.yml

@@ -4,8 +4,8 @@ channels:
   - nvidia
   - rapidsai
 dependencies:
-  - cuda-version=11.8
-  - cudf=24.12
+  - cuda-version=12.8
+  - cudf=25.02
   - libcusolver
   - libcusparse
   - matplotlib

conda/environments/triton_test.yml

@@ -5,10 +5,10 @@ channels:
   - rapidsai
 dependencies:
   - aws-sdk-cpp
-  - clang-tools=11.1.0
-  - cuda-version=11.8
-  - cudf=24.12
-  - cuml=24.12
+  - clang-tools=19.1.7
+  - cuda-version=12.8
+  - cudf=25.02
+  - cuml=25.02
   - flake8
   - hypothesis
   - lightgbm

conda/environments/triton_test_no_client.yml

@@ -5,10 +5,10 @@ channels:
   - rapidsai
 dependencies:
   - aws-sdk-cpp
-  - clang-tools=11.1.0
-  - cuda-version=11.8
-  - cudf=24.12
-  - cuml=24.12
+  - clang-tools=19.1.7
+  - cuda-version=12.8
+  - cudf=25.02
+  - cuml=25.02
   - flake8
   - hypothesis
   - lightgbm

ops/Dockerfile

@@ -3,7 +3,7 @@
 # Arguments for controlling build details
 ###########################################################################################
 # Version of Triton to use
-ARG TRITON_VERSION=25.01
+ARG TRITON_VERSION=25.02
 # Base container image
 ARG BASE_IMAGE=nvcr.io/nvidia/tritonserver:${TRITON_VERSION}-py3
 # Whether or not to enable GPU build
@@ -63,7 +63,7 @@ RUN conda run --no-capture-output -n triton_test \
 
 FROM wheel-install-${USE_CLIENT_WHEEL} as conda-test
 RUN conda run --no-capture-output -n triton_test \
-    pip install git+https://github.com/rapidsai/rapids-triton.git@branch-24.10#subdirectory=python
+    pip install git+https://github.com/rapidsai/rapids-triton.git@branch-25.02#subdirectory=python
 RUN conda-pack --ignore-missing-files -n triton_test -o /tmp/env.tar \
  && mkdir /conda/test/ \
  && cd /conda/test/ \

scripts/environment.yml

@@ -4,8 +4,8 @@ channels:
   - nvidia
   - rapidsai
 dependencies:
-  - cuda-version=11.8
-  - cuml=24.12
+  - cuda-version=12.8
+  - cuml=25.02
   - python
   - scikit-learn>=1.5
   - treelite>=4.3