Prep for multi image Batch-Shipyard docker repo

cascade/Dockerfile

@@ -20,7 +20,7 @@ RUN apk update \
     && tar zxvpf libtorrent-${libtorrent_version}.tar.gz \
     && cd libtorrent-rasterbar-${libtorrent_version} \
     && ./configure --prefix=/usr --enable-debug=no --enable-python-binding --with-boost-system=boost_system \
-    && make -j"$(nproc)" install \
+    && make -j4 install \
     && ldconfig /usr/lib \
     && cd .. \
     && rm -rf libtorrent-rasterbar-${libtorrent_version} \

recipes/MXNet-GPU/README.md

@@ -1,4 +1,4 @@
-# MXNet-CPU
+# MXNet-GPU
 This recipe shows how to run [MXNet](http://mxnet.io/) on GPUs on one or
 N-series Azure VM instances in an Azure Batch compute pool.
 

recipes/MXNet-GPU/docker/README.md

@@ -1,4 +1,4 @@
-# Dockerfile for MXNet-CPU
+# Dockerfile for MXNet-GPU
 This image can be found on [Docker Hub](https://hub.docker.com/r/alfpark/mxnet/).
 
 You must agree to the following licenses prior to use:

scripts/shipyard_nodeprep.sh

@@ -574,7 +574,7 @@ EOF
         -v $AZ_BATCH_NODE_ROOT_DIR:$AZ_BATCH_NODE_ROOT_DIR \
         -w $AZ_BATCH_TASK_WORKING_DIR \
         -p 6881-6891:6881-6891 -p 6881-6891:6881-6891/udp \
-        alfpark/batch-shipyard
+        alfpark/batch-shipyard:cascade-latest
 else
     # backfill node prep start
     if [ ! -z ${SHIPYARD_TIMING+x} ]; then