[GitHub] [incubator-mxnet] DickJC123 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>]

DickJC123 commented on a change in pull request #19149:
URL: https://github.com/apache/incubator-mxnet/pull/19149#discussion_r489085353



##########
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                              \
+        USE_CUDA=1                                \
+        USE_CUDA_PATH=/usr/local/cuda             \
+        USE_CUDNN=1                               \
+        USE_TVM_OP=0                              \
+        USE_CPP_PACKAGE=1                         \
+        USE_DIST_KVSTORE=1                        \
+        CUDA_ARCH="$CI_CUDA_COMPUTE_CAPABILITIES" \

Review comment:
       I notice that earlier in this file, we have the lines
   ```
   CI_CUDA_COMPUTE_CAPABILITIES="-gencode=arch=compute_52,code=sm_52 -gencode=arch=compute_70,code=sm_70"
   CI_CMAKE_CUDA_ARCH="5.2 7.0"
   ```
   Not sure what we're trying to prove exactly with this PR, but do we gain any confidence by adding arch 80, even though there are no A100's in the CI?




----------------------------------------------------------------
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