Provide 24.11 version bumps (#413)

* Provide 24.11 version bumps

* geventhttpclient 2.0.2 must be built from source; no binary wheel for Python 3.12

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    24.08
+find_and_configure_cuml(VERSION    24.10
                         FORK       rapidsai
-                        PINNED_TAG branch-24.08
+                        PINNED_TAG feature/fil-backend-r24.11
                         USE_TREELITE_STATIC ${TRITON_FIL_USE_TREELITE_STATIC}
                         )

cmake/thirdparty/get_treelite.cmake

@@ -78,6 +78,6 @@ function(find_and_configure_treelite)
     rapids_export_find_package_root(BUILD Treelite [=[${CMAKE_CURRENT_LIST_DIR}]=] EXPORT_SET cuml-exports)
 endfunction()
 
-find_and_configure_treelite(VERSION     4.3.0
-                        PINNED_TAG  575e4208f2b18e40d818c338ecb95d7a26e69aab
+find_and_configure_treelite(VERSION     4.4.0
+                        PINNED_TAG  da70beb846dc5bc235643684b1b6176585dd094a
                         BUILD_STATIC_LIBS ${TRITON_FIL_USE_TREELITE_STATIC})

conda/environments/triton_benchmark.yml

@@ -5,7 +5,7 @@ channels:
   - rapidsai
 dependencies:
   - cuda-version=11.8
-  - cudf=24.08
+  - cudf=24.10
   - libcusolver
   - libcusparse
   - matplotlib

conda/environments/triton_test.yml

@@ -7,8 +7,8 @@ dependencies:
   - aws-sdk-cpp
   - clang-tools=11.1.0
   - cuda-version=11.8
-  - cudf=24.08
-  - cuml=24.08
+  - cudf=24.10
+  - cuml=24.10
   - flake8
   - hypothesis
   - lightgbm

conda/environments/triton_test_no_client.yml

@@ -7,8 +7,8 @@ dependencies:
   - aws-sdk-cpp
   - clang-tools=11.1.0
   - cuda-version=11.8
-  - cudf=24.08
-  - cuml=24.08
+  - cudf=24.10
+  - cuml=24.10
   - flake8
   - hypothesis
   - lightgbm

ops/Dockerfile

@@ -3,7 +3,7 @@
 # Arguments for controlling build details
 ###########################################################################################
 # Version of Triton to use
-ARG TRITON_VERSION=24.09
+ARG TRITON_VERSION=24.10
 # Base container image
 ARG BASE_IMAGE=nvcr.io/nvidia/tritonserver:${TRITON_VERSION}-py3
 # Whether or not to enable GPU build
@@ -17,7 +17,7 @@ ARG SDK_IMAGE=nvcr.io/nvidia/tritonserver:${TRITON_VERSION}-py3-sdk
 # Whether or not to use backend library prebuilt on host
 ARG USE_HOST_LIB=0
 
-FROM condaforge/mambaforge as conda-base
+FROM condaforge/miniforge3 as conda-base
 COPY ./ops/gpuci_conda_retry /usr/bin/gpuci_conda_retry
 COPY ./ops/gpuci_mamba_retry /usr/bin/gpuci_mamba_retry
 RUN chmod +x /usr/bin/gpuci_conda_retry /usr/bin/gpuci_mamba_retry
@@ -44,7 +44,14 @@ RUN gpuci_mamba_retry env update -f /environment.yml \
     && rm /environment.yml
 
 FROM base-test-install as wheel-install-0
-RUN conda run --no-capture-output -n triton_test pip install tritonclient[all]
+RUN apt-get update \
+    && apt-get install --no-install-recommends -y \
+      build-essential \
+      ca-certificates \
+      git \
+    && apt-get clean \
+    && rm -rf /var/lib/apt/lists/* \
+    && conda run --no-capture-output -n triton_test pip install tritonclient[all]
 
 FROM ${SDK_IMAGE} as sdk-image
 

scripts/environment.yml

@@ -5,7 +5,7 @@ channels:
   - rapidsai
 dependencies:
   - cuda-version=11.8
-  - cuml=24.08
+  - cuml=24.10
   - python
   - scikit-learn>=1.5
   - treelite>=4.3