container_deep_learning_gpu_centos7_x86_64

Container recipes for Deep Learning with GPU for architecture x86_64 based on a CentOS 7. In the specific:

• CentOS 7 with cuda library (10.0-devel-centos7)
• GNU compiler 7
• Python 3.6
• OpenMPI 2.1.1 (compiled with support for psm2, pmix, verbs)
• Tensorflow 1.14.0 GPU (pip)
• Py-Torch 1.4.0 GPU (pip)
• Torchvision 0.5.0 CPU (pip)
• MxNet 1.5.1 CPU (pip)
• Horovod 0.19.1 (compiled with Tensorflow, Pytorch, MxNet)

This recipe works on in the Cineca cluster (arch x86_64):

• Galileo