Caffe is a Deep Learning framework developed by the Berkeley Vision and Learning Center (BVLC) and by community contributors. Yangqing Jia created the project during his PhD at UC Berkeley. The Caffe BLVC is a single GPU and single node framework.
Inspur further developed the Caffe and added clustering capabilities to the framework and created Caffe-MPI, a parallel GPU cluster framework to speed up Deep Learning.
Both versions are Open Source and are available for download at GitHub:
Inspur Caffe MPI: https://github.com/Caffe-MPI/Caffe-MPI.github.io