Bump dependency versions for 23.07 (#360)

* Bump dependency versions for 23.07

* Update Python version for RAPIDS 23.06

CMakeLists.txt

@@ -24,8 +24,8 @@ set(TRITON_CORE_REPO_TAG "main" CACHE STRING "Tag for triton-inference-server/co
 set(TRITON_BACKEND_REPO_TAG "main" CACHE STRING "Tag for triton-inference-server/backend repo")
 
 if(TRITON_FIL_DOCKER_BUILD)
-  set(TRITON_BUILD_CONTAINER "nvcr.io/nvidia/tritonserver:23.05-py3" CACHE STRING "Build image for Dockerized builds")
-  set(TRITON_BUILD_CONTAINER_VERSION "23.05" CACHE STRING "Triton version for Dockerized builds")
+  set(TRITON_BUILD_CONTAINER "nvcr.io/nvidia/tritonserver:23.06-py3" CACHE STRING "Build image for Dockerized builds")
+  set(TRITON_BUILD_CONTAINER_VERSION "23.06" CACHE STRING "Triton version for Dockerized builds")
 
   add_custom_command(
     OUTPUT fil/libtriton_fil.so $<$<BOOL:${TRITON_ENABLE_GPU}>:fil/libcuml++.so>
@@ -70,7 +70,7 @@ else()
   ##############################################################################
   # - Prepare rapids-cmake -----------------------------------------------------
   file(DOWNLOAD
-    https://raw.githubusercontent.com/rapidsai/rapids-cmake/branch-23.04/RAPIDS.cmake
+    https://raw.githubusercontent.com/rapidsai/rapids-cmake/branch-23.06/RAPIDS.cmake
       ${CMAKE_BINARY_DIR}/RAPIDS.cmake)
   include(${CMAKE_BINARY_DIR}/RAPIDS.cmake)
   include(rapids-cmake)
@@ -89,7 +89,7 @@ else()
   option(DISABLE_DEPRECATION_WARNINGS "Disable depreaction warnings " ON)
   option(NVTX "Enable nvtx markers" OFF)
   set(BACKEND_FOLDER "/opt/tritonserver/backends" CACHE STRING "Triton backend folder path")
-  set(RAPIDS_DEPENDENCIES_VERSION "23.04" CACHE STRING "RAPIDS projects dependencies version")
+  set(RAPIDS_DEPENDENCIES_VERSION "23.06" CACHE STRING "RAPIDS projects dependencies version")
   option(TRITON_FIL_USE_TREELITE_STATIC "Link Treelite statically in libtriton_fil.so and cuml++.so" ON)
 
 

README.md

@@ -152,7 +152,7 @@ dynamic_batching {}
 ```
 docker run -p 8000:8000 -p 8001:8001 --gpus all \
   -v ${PWD}/model_repository:/models \
-  nvcr.io/nvidia/tritonserver:23.05-py3 \
+  nvcr.io/nvidia/tritonserver:23.06-py3 \
   tritonserver --model-repository=/models
 ```
 

build.sh

@@ -168,7 +168,7 @@ DOCKER_ARGS="$DOCKER_ARGS --build-arg TRITON_ENABLE_GPU=${TRITON_ENABLE_GPU}"
 
 if [ -z $RAPIDS_VERSION ]
 then
-  RAPIDS_VERSION=23.04
+  RAPIDS_VERSION=23.06
 else
   DOCKER_ARGS="$DOCKER_ARGS --build-arg RAPIDS_DEPENDENCIES_VERSION=${RAPIDS_VERSION}"
 fi
@@ -190,7 +190,7 @@ if [ -z $TRITON_VERSION ] && [ $HOST_BUILD -eq 1 ]
 then
   # Must use a version compatible with a released backend image in order to
   # test a host build, so default to latest release branch rather than main
-  TRITON_VERSION=23.05
+  TRITON_VERSION=23.06
 fi
 
 if [ ! -z $TRITON_VERSION ]

cmake/thirdparty/get_cuml.cmake

@@ -57,6 +57,6 @@ endfunction()
 # CPM_raft_SOURCE=/path/to/local/raft
 find_and_configure_cuml(VERSION    ${RAPIDS_TRITON_MIN_VERSION_rapids_projects}
                         FORK       rapidsai
-                        PINNED_TAG branch-23.04
+                        PINNED_TAG branch-23.06
                         USE_TREELITE_STATIC ${TRITON_FIL_USE_TREELITE_STATIC}
                         )

conda/environments/triton_benchmark.yml

@@ -5,14 +5,14 @@ channels:
   - rapidsai
 dependencies:
   - cudatoolkit=11.5
-  - cudf=23.04
+  - cudf=23.06
   - libcusolver<=11.4.1.48
   - libcusparse<=12.0
   - matplotlib
   - pip
-  - python=3.8
+  - python=3.10
   - scipy
   - pip:
       - tritonclient[all]
       - protobuf==3.20.1
-      - git+https://github.com/rapidsai/rapids-triton.git@branch-23.04#subdirectory=python
+      - git+https://github.com/rapidsai/rapids-triton.git@branch-23.06#subdirectory=python

conda/environments/triton_test.yml

@@ -7,19 +7,19 @@ dependencies:
   - aws-sdk-cpp
   - clang-tools=11.1.0
   - cudatoolkit=11.5
-  - cudf=23.04
-  - cuml=23.04
+  - cudf=23.06
+  - cuml=23.06
   - flake8
   - hypothesis<6.46.8
   - lightgbm
   - matplotlib
   - pip
   - pytest
-  - python=3.8
+  - python=3.10
   - rapidsai::xgboost>=1.7
   - scikit-learn=1.2.0
   - treelite
   - pip:
       - tritonclient[all]
       - protobuf==3.20.1
-      - git+https://github.com/rapidsai/rapids-triton.git@branch-23.04#subdirectory=python
+      - git+https://github.com/rapidsai/rapids-triton.git@branch-23.06#subdirectory=python

conda/environments/triton_test_no_client.yml

@@ -7,14 +7,14 @@ dependencies:
   - aws-sdk-cpp
   - clang-tools=11.1.0
   - cudatoolkit=11.5
-  - cudf=23.04
-  - cuml=23.04
+  - cudf=23.06
+  - cuml=23.06
   - flake8
   - hypothesis<6.46.8
   - lightgbm
   - pip
   - pytest
-  - python=3.8
+  - python=3.10
   - python-rapidjson
   - rapidsai::xgboost>=1.7
   - scikit-learn=1.2.0

ops/Dockerfile

@@ -3,7 +3,7 @@
 # Arguments for controlling build details
 ###########################################################################################
 # Version of Triton to use
-ARG TRITON_VERSION=23.05
+ARG TRITON_VERSION=23.06
 # Base container image
 ARG BASE_IMAGE=nvcr.io/nvidia/tritonserver:${TRITON_VERSION}-py3
 # Whether or not to enable GPU build
@@ -135,7 +135,7 @@ ENV TRITON_ENABLE_GPU=$TRITON_ENABLE_GPU
 ARG TRITON_ENABLE_STATS=ON
 ENV TRITON_ENABLE_GPU=$TRITON_ENABLE_GPU
 
-ARG RAPIDS_DEPENDENCIES_VERSION=23.04
+ARG RAPIDS_DEPENDENCIES_VERSION=23.06
 ENV RAPIDS_DEPENDENCIES_VERSION=$RAPIDS_DEPENDENCIES_VERSION
 
 ARG TRITON_FIL_USE_TREELITE_STATIC=ON

scripts/environment.yml

@@ -5,7 +5,7 @@ channels:
   - rapidsai
 dependencies:
   - cudatoolkit=11.5
-  - cuml=23.04
+  - cuml=23.06
   - python=3.8
   - scikit-learn
   - treelite