[GitHub] [incubator-mxnet] ChaiBapchya commented on a change in pull request #19149: [v1.x] Add new CI pipeline for building and testing with cuda 11.0.

[GitBox <gi...@apache.org>]

ChaiBapchya commented on a change in pull request #19149:
URL: https://github.com/apache/incubator-mxnet/pull/19149#discussion_r489057391



##########
File path: ci/docker/Dockerfile.build.ubuntu_build_cuda110
##########
@@ -0,0 +1,46 @@
+# -*- mode: dockerfile -*-
+# Licensed to the Apache Software Foundation (ASF) under one
+# or more contributor license agreements.  See the NOTICE file
+# distributed with this work for additional information
+# regarding copyright ownership.  The ASF licenses this file
+# to you under the Apache License, Version 2.0 (the
+# "License"); you may not use this file except in compliance
+# with the License.  You may obtain a copy of the License at
+#
+#   http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing,
+# software distributed under the License is distributed on an
+# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+# KIND, either express or implied.  See the License for the
+# specific language governing permissions and limitations
+# under the License.
+#
+# Dockerfile to build MXNet on Ubuntu 16.04 for GPU but on
+# a CPU-only instance. This restriction is caused by the CPP-
+# package generation, requiring the actual CUDA library to be
+# present
+
+FROM nvidia/cuda:11.0-cudnn8-devel-ubuntu16.04
+
+WORKDIR /work/deps
+
+COPY install/ubuntu_core.sh /work/
+RUN /work/ubuntu_core.sh
+COPY install/deb_ubuntu_ccache.sh /work/
+RUN /work/deb_ubuntu_ccache.sh
+COPY install/ubuntu_python.sh /work/
+COPY install/requirements /work/
+RUN /work/ubuntu_python.sh
+
+# Keep this at the end since this command is not cachable
+ARG USER_ID=0
+ARG GROUP_ID=0
+COPY install/ubuntu_adduser.sh /work/
+RUN /work/ubuntu_adduser.sh
+
+COPY runtime_functions.sh /work/
+
+WORKDIR /work/mxnet
+ENV LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/usr/local/cuda/compat

Review comment:
       I remember facing issues with cuda/compat directory while migrating from G3 to G4. Do we need this? @ptrendx @leezu confirm.
   It was only TVM OP which needed that right? And since we are disabling it we shouldn't need compat dir.
   
   Correct me if wrong.

##########
File path: ci/docker/runtime_functions.sh
##########
@@ -806,6 +806,24 @@ build_ubuntu_gpu_cuda101_cudnn7() {
     make cython PYTHON=python3
 }
 
+build_ubuntu_gpu_cuda110_cudnn8() {
+    set -ex
+    build_ccache_wrappers
+    make \
+        USE_BLAS=openblas                         \
+        USE_MKLDNN=0                              \

Review comment:
       MKLDNN is ON by default right?




----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org