Import system images

System-level container images are used by all users. You can obtain images from Lenovo salesperson or import images into LiCO as system-level container images. This section describes how to create and import system-level container images.

Image bootstrap files

Download image_bootstrap.zip from https://hpc.lenovo.com/lico/downloads/6.4/images/host/image_bootstrap.zip, image_bootstrap.zip includes the following singularity bootstrap(definition) files, then we can use it to create images.

File name	Framework	CPU/GPU	Comments
caffe-1.0-cpu	Caffe	CPU
caffe-1.0-gpu-cuda92	Caffe	CUDA 9.2	Supports P100 and V100 Caffe does not support CUDA 9.0 officially
NVCaffe-0.17.3-gpu-cuda102	Caffe	CUDA 10.2	Supports P100 and V100
chainer-6.7.0-gpu-cuda101	Chainer	CUDA 10.1	Supports P100, V100, RTX5000, RTX8000 and T4
intel-caffe-1.1.6-cpu	Intel-caffe	CPU
intel-python-3.7	Other	CPU
intel-pytorch-1.7.0-cpu	PyTorch	CPU
intel-tensorflow-1.15.2-cpu	TensorFlow	CPU
intel-tensorflow-2.3.0-cpu	TensorFlow	CPU
jupyter-intel-optimized-pytorch	Jupyter	CPU
jupyter-intel-optimized-tensorflow	Jupyter	CPU
jupyter-default	Jupyter	CPU CUDA 11.2	Supports P100, V100, RTX5000, RTX8000, T4 and A100
jupyter-py37	Jupyter	CPU CUDA 11.0	Supports P100, V100, RTX5000, RTX8000, T4 and A100
jupyter-py38	Jupyter	CPU CUDA 11.0	Supports P100, V100, RTX5000, RTX8000, T4 and A100
jupyter-intel-py37	Jupyter	CPU
letrain-1.6.0-gpu-cuda110	LeTrain	CPU CUDA 11.0	Supports P100, V100, RTX5000, RTX8000, T4 and A100
mxnet-1.5.0-cpu-mkl	Mxnet	CPU
mxnet-1.5.0-gpu-mkl-cuda101	Mxnet	CUDA 10.1	Supports P100, V100, RTX5000, RTX8000 and T4
neon-2.6-cpu	Neon	CPU
pytorch-1.11.0-cuda115	PyTorch	CUDA 11.5	Supports P100, V100, RTX5000, RTX8000, T4 and A100
rstudio	RStudio	CPU CUDA 11.2	Supports P100, V100, RTX5000, RTX8000, T4 and A100
scikit-single-cpu	Scikit	CPU
tensorflow-1.15.3-cuda110	TensorFlow	CPU CUDA 11.0	Supports P100, V100, RTX5000, RTX8000, T4 and A100
tensorflow-1.15.3-cuda110-hbase	TensorFlow	CPU CUDA 11.0	Supports HBase Supports P100, V100, RTX5000, RTX8000, T4 and A100
tensorflow-1.15.3-cuda110-keras	TensorFlow	CPU CUDA 11.0	Supports Keras(2.2.4) Supports P100, V100, RTX5000, RTX8000, T4 and A100
tensorflow-1.15.3-cuda110-mongodb	TensorFlow	CPU CUDA 11.0	Supports MongoDB Supports P100, V100, RTX5000, RTX8000, T4 and A100
tensorflow-1.15.2-mkl	TensorFlow	CPU
tensorflow-2.5.0-cuda114	TensorFlow	CPU CUDA 11.4	Supports P100, V100, RTX5000, RTX8000, T4 and A100
tensorrt-7.1.3.4-cuda110	TensorRT	CUDA 11.0	Supports P100, V100 and A100
cvat	Other	CPU

Create images

Step 1. Prepare a build node with a minimum storage of 100 GB.

Notes:

This node should have the same OS version and singularity version with the nodes in the cluster.
If you want to create GPU images, this node should have the same GPU and GPU driver as nodes in the cluster.
To use images related to the MKL function, ensure that these images are created on a machine with the same CPU architecture.

Step 2. To the build node, ensure that squashfs-tools, libarchive, and make are installed.

Step 3. To the build node, upload the compressed image bootstrap file you obtained which named image_bootstrap.zip. For example, upload the compressed package to the new directory /opt/images. If the new directory cannot be found, create it manually. Note that both this new directory and /var/tmp cannot be an NFS mount.

Step 4. To the build node, run the following commands to extract the compressed package.

 
cd /opt/images
unzip image_bootstrap.zip

Step 5. To the build node, do one of the following to create image. The created image file is in the dist folder of the current directory.

Run the following commands to create all images at once:

 
xxxxxxxxxx
cd /opt/images/image_bootstrap
make all

Run the following commands to create a group images at once:

 
xxxxxxxxxx
cd /opt/images/image_bootstrap
make caffe
make cvat
make intel-caffe
make intel-python
make tensorflow
make mxnet
make neon
make chainer
make letrain
make jupyter
make pytorch
make rstudio
make scikit
make tensorrt
make intel-pytorch
make intel-tensorflow

Notes: Check the network when one of the following errors is displayed:

ERROR MANIFEST_UNKNOWN: manifest unknown
ReadTimeoutError: HTTPSConnectionPool(host='files.pythonhosted.org', port=443): Read timed out.
urllib2.URLError:

Import images into LiCO as system-level images

Step 1. Copy the created images to the management node.

For example, copy the images to directory /opt/images. Note that both this directory and /var/tmp cannot be an NFS mount

Step 2. Run the following commands to import images into LiCO:

 
x
cd /opt/images
lico import_image caffe-cpu caffe-1.0-cpu.image caffe
lico import_image caffe-gpu caffe-1.0-gpu-cuda92.image caffe
lico import_image NVCaffe NVCaffe-0.17.3-gpu-cuda102.image caffe
lico import_image intel-caffe intel-caffe-1.1.6-cpu.image intel-caffe
lico import_image intel-python intel-python-3.7.image other
lico import_image tensorflow tensorflow-1.15.3-cuda110.image tensorflow
lico import_image tensorflow-mkl tensorflow-1.15.2-mkl.image tensorflow
lico import_image tensorflow-hbase tensorflow-1.15.3-cuda110-hbase.image tensorflow
lico import_image tensorflow-keras tensorflow-1.15.3-cuda110-keras.image tensorflow
lico import_image tensorflow-mongodb tensorflow-1.15.3-cuda110-mongodb.image tensorflow
lico import_image tensorflow2 tensorflow-2.5.0-cuda114.image tensorflow2
lico import_image intel-tensorflow intel-tensorflow-1.15.2-cpu.image tensorflow
lico import_image intel-tensorflow2 intel-tensorflow-2.3.0-cpu.image tensorflow2
lico import_image mxnet-cpu mxnet-1.5.0-cpu-mkl.image mxnet
lico import_image mxnet-gpu mxnet-1.5.0-gpu-mkl-cuda100.image mxnet
lico import_image neon neon-2.6-cpu.image neon
lico import_image chainer-gpu chainer-6.7.0-gpu-cuda101.image chainer
lico import_image letrain letrain-1.6.0-cuda110.image letrain
lico import_image jupyter-default jupyter-default.image jupyter -t py38 -t cpu -t gpu
lico import_image jupyter-py37 jupyter-py37.image jupyter -t py37 -t cpu -t gpu
lico import_image jupyter-py38 jupyter-py38.image jupyter -t py38 -t cpu -t gpu
lico import_image jupyter-intel-py37 jupyter-intel-py37.image jupyter -t intel_py37 -t cpu
lico import_image pytorch pytorch-1.11.0-cuda115.image pytorch
lico import_image intel-pytorch intel-pytorch-1.7.0-cpu.image pytorch
lico import_image rstudio rstudio.image rstudio -t cpu -t gpu
lico import_image scikit-cpu scikit-single-cpu.image scikit
lico import_image tensorrt tensorrt-7.1.3.4.image tensorrt -t tensorrt
lico import_image cvat cvat-1.7.0.image other
# If you use the intel openapi module, please import the following image
lico import_image jupyter-intel-optimized-pytorch \
jupyter-intel-optimized-pytorch.image jupyter -t pytorch
lico import_image jupyter-intel-optimized-tensorflow \
jupyter-intel-optimized-tensorflow.image jupyter -t tensorflow2