Release 25.08.0 (#978)

Merge branch-25.08 into main

Release notes:
- Updates RAPIDS dependencies to 25.08
- Bug fixes in UMAP (Hellinger distance) and RandomForest
- Drops support for cuda 11.

Known issues:
- CrossValidator for RandomForest over Spark Connect will fail in Spark
4.0. Fix pending in Spark 4.1



Note: merge this PR with **Create a merge commit to merge**

Author: YanxuanLiu

ci/Dockerfile

@@ -14,7 +14,7 @@
 # limitations under the License.
 #
 
-ARG CUDA_VERSION=11.8.0
+ARG CUDA_VERSION=12.0.1
 FROM nvidia/cuda:${CUDA_VERSION}-devel-ubuntu22.04
 
 # ubuntu22
@@ -37,6 +37,7 @@ ENV JAVA_HOME /usr/lib/jvm/java-1.17.0-openjdk-amd64
 
 # Install conda
 ENV PATH="/root/miniconda3/bin:${PATH}"
+ENV CONDA_PLUGINS_AUTO_ACCEPT_TOS="yes"
 RUN wget --quiet https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh \
     && mkdir /root/.conda \
     && bash Miniconda3-latest-Linux-x86_64.sh -b \
@@ -46,6 +47,6 @@ RUN wget --quiet https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86
     && conda config --set solver libmamba
 
 # install cuML
-ARG CUML_VER=25.06
-RUN conda install -y -c rapidsai -c conda-forge -c nvidia cuml=$CUML_VER cuvs=$CUML_VER python=3.10 cuda-version=11.8 numpy~=1.0 \
+ARG CUML_VER=25.08
+RUN conda install -y -c rapidsai -c conda-forge -c nvidia cuml=$CUML_VER cuvs=$CUML_VER python=3.10 cuda-version=12.0 numpy~=1.0 \
     && conda clean --all -f -y

ci/Jenkinsfile.premerge

@@ -30,7 +30,7 @@ import ipp.blossom.*
 
 def githubHelper // blossom github helper
 def TEMP_IMAGE_BUILD = true
-def IMAGE_PREMERGE = "${common.ARTIFACTORY_NAME}/sw-spark-docker/rapids:ml-ubuntu22-cuda11.8.0-py310"
+def IMAGE_PREMERGE = "${common.ARTIFACTORY_NAME}/sw-spark-docker/rapids:ml-ubuntu22-cuda12.0.1-py310"
 def cpuImage = pod.getCPUYAML("${common.ARTIFACTORY_NAME}/sw-spark-docker/spark:rapids-databricks") // tooling image
 def PREMERGE_DOCKERFILE = 'ci/Dockerfile'
 def PREMERGE_TAG

docker/Dockerfile

@@ -35,7 +35,7 @@
 #
 ###
 
-ARG CUDA_VERSION=11.5.2
+ARG CUDA_VERSION=12.0.1
 FROM nvidia/cuda:${CUDA_VERSION}-devel-ubuntu20.04
 
 # ubuntu22
@@ -91,6 +91,7 @@ RUN wget --quiet \
     && mkdir /root/.conda \
     && bash Miniconda3-py38_4.10.3-Linux-x86_64.sh -b \
     && rm -f Miniconda3-py38_4.10.3-Linux-x86_64.sh \
+    && conda tos accept --override-channels -c conda-forge -c defaults \
     && conda init
 
 # install cuDF dependency, Fall back to use cudf 22.04 due to issue:

docker/Dockerfile.pip

@@ -14,11 +14,11 @@
 # limitations under the License.
 #
 
-ARG CUDA_VERSION=11.8.0
+ARG CUDA_VERSION=12.0.1
 FROM nvidia/cuda:${CUDA_VERSION}-devel-ubuntu22.04
 
 ARG PYSPARK_VERSION=3.3.1
-ARG RAPIDS_VERSION=25.6.0
+ARG RAPIDS_VERSION=25.8.0
 ARG ARCH=amd64
 #ARG ARCH=arm64
 
@@ -47,9 +47,9 @@ RUN apt-get update -y \
 # install RAPIDS
 # using ~= pulls in micro version patches
 RUN pip install --no-cache-dir \
-    cudf-cu11~=${RAPIDS_VERSION} \
-    cuml-cu11~=${RAPIDS_VERSION} \
-    cuvs-cu11~=${RAPIDS_VERSION} \
+    cudf-cu12~=${RAPIDS_VERSION} \
+    cuml-cu12~=${RAPIDS_VERSION} \
+    cuvs-cu12~=${RAPIDS_VERSION} \
     numpy~=1.0 \
     --extra-index-url=https://pypi.nvidia.com
 

docker/Dockerfile.python

@@ -14,10 +14,10 @@
 # limitations under the License.
 #
 
-ARG CUDA_VERSION=11.8.0
+ARG CUDA_VERSION=12.0.1
 FROM nvidia/cuda:${CUDA_VERSION}-devel-ubuntu22.04
 
-ARG CUML_VERSION=25.06
+ARG CUML_VERSION=25.08
 
 # ubuntu22
 RUN sed -i -e 's|http://archive.ubuntu.com/ubuntu|https://archive.ubuntu.com/ubuntu|g' \
@@ -42,11 +42,12 @@ RUN wget --quiet https://repo.anaconda.com/miniconda/Miniconda3-py38_4.10.3-Linu
     && mkdir /root/.conda \
     && bash Miniconda3-py38_4.10.3-Linux-x86_64.sh -b \
     && rm -f Miniconda3-py38_4.10.3-Linux-x86_64.sh \
+    && conda tos accept --override-channels -c conda-forge -c defaults \
     && conda init
 
 # install cuML
 
-RUN conda install -y -c rapidsai -c conda-forge -c nvidia python=3.10 cuda-version=11.8 cuml=$CUML_VERSION numpy~=1.0 \
+RUN conda install -y -c rapidsai -c conda-forge -c nvidia python=3.10 cuda-version=12.0 cuml=$CUML_VERSION numpy~=1.0 \
     && conda clean --all -f -y
 
 # install python dependencies

docs/site/FAQ.md

@@ -14,3 +14,9 @@ Apache Spark version 3.3.1 or higher.
 ### What versions of Python are supported
 
 Python 3.10 or higher.
+
+### How do I fix the "java.lang.IllegalArgumentException: valueCount must be >= 0" error?
+
+This error occurs when the product of Arrow batch size and row dimension exceeds 2,147,483,647 (INT32_MAX), typically with very wide datasets (many features per row), causing Arrow serialization to fail. For example, if you set `max_records_per_batch = 10000` and your data has `row_dimension = 300000` (i.e., 300,000 features per row), then `10000 × 300000 = 3,000,000,000`, which exceeds the Arrow limit of 2,147,483,647 (INT32_MAX) and will cause this error.
+
+Be aware that some Spark Rapids ML algorithms (such as NearestNeighbors) may convert sparse vectors to dense format internally if the underlying cuML algorithm does not support sparse input. This conversion can significantly increase memory usage, especially with wide datasets, and may make the Arrow size limit error more likely. To mitigate this, lower the value of `spark.sql.execution.arrow.maxRecordsPerBatch` (for example, to 5,000 or less) so that the product of the batch size and the number of elements per row stays within Arrow's maximum allowed size.

docs/site/compatibility.md

@@ -31,7 +31,7 @@ The following table shows the currently supported algorithms.  The goal is to ex
 
 | Spark Rapids ML | CUDA  | Spark  | Python |
 | :-------------- | :---- | :----- | :----- |
-| 1.0.0           | 11.4+ | 3.3+   | 3.10+  |
+| 1.0.0           | 12.0+ | 3.3+   | 3.10+  |
 
 
 ## Single vs Double precision inputs

docs/source/conf.py

@@ -23,7 +23,7 @@
 project = 'spark-rapids-ml'
 copyright = '2025, NVIDIA'
 author = 'NVIDIA'
-release = '25.06.0'
+release = '25.08.0'
 
 # -- General configuration ---------------------------------------------------
 # https://www.sphinx-doc.org/en/master/usage/configuration.html#general-configuration

jvm/README.md

@@ -50,10 +50,10 @@ including setting up the server and running client-side tests.
 To start the Spark Connect server with Spark Rapids ML support, follow these steps:
 
 ```shell
-conda activate rapids-25.06  # from spark-rapids-ml installation
+conda activate rapids-25.08  # from spark-rapids-ml installation
 export SPARK_HOME=<directory where spark was installed above>
 export PYSPARK_PYTHON=$(which python)
-export PLUGIN_JAR=$(pip show spark-rapids-ml | grep Location: | cut -d ' ' -f 2 )/spark_rapids_ml/jars/com.nvidia.rapids.ml-25.06.0.jar
+export PLUGIN_JAR=$(pip show spark-rapids-ml | grep Location: | cut -d ' ' -f 2 )/spark_rapids_ml/jars/com.nvidia.rapids.ml-25.08.0.jar
 $SPARK_HOME/sbin/start-connect-server.sh --master local[*] \
   --jars $PLUGIN_JAR \
   --conf spark.driver.memory=20G
@@ -107,7 +107,7 @@ mvn clean package -DskipTests
 if you would like to compile the plugin and run the unit tests, install `spark-rapids-ml` python package and its dependencies per the above instructions and run the following command:
 
 ``` shell
-conda activate rapids-25.06
+conda activate rapids-25.08
 export PYSPARK_PYTHON=$(which python)
 mvn clean package
 ```

jvm/pom.xml

@@ -21,7 +21,7 @@
 
     <groupId>com.nvidia.rapids</groupId>
     <artifactId>ml</artifactId>
-    <version>25.06.0</version>
+    <version>25.08.0</version>
     <packaging>jar</packaging>
 
     <properties>