Add use_rmm flag to global configuration (#6656)

* Ensure RMM is 0.18 or later

* Add use_rmm flag to global configuration

* Modify XGBCachingDeviceAllocatorImpl to skip CUB when use_rmm=True

* Update the demo

* [CI] Pin NumPy to 1.19.4, since NumPy 1.19.5 doesn't work with latest Shap
This commit is contained in:
Philip Hyunsu Cho
2021-03-09 14:53:05 -08:00
committed by GitHub
parent e4894111ba
commit 366f3cb9d8
12 changed files with 117 additions and 20 deletions

View File

@@ -28,7 +28,7 @@ ENV PATH=/opt/python/bin:$PATH
# Create new Conda environment with RMM
RUN \
conda create -n gpu_test -c nvidia -c rapidsai-nightly -c rapidsai -c conda-forge -c defaults \
conda create -n gpu_test -c nvidia -c rapidsai -c conda-forge -c defaults \
python=3.7 rmm=0.18* cudatoolkit=$CUDA_VERSION_ARG
ENV GOSU_VERSION 1.10