Fraunhofer ITWM – CC HPC Distributed Deep Learning

Features:

 Better scalability using asynchronous PGAS programming of optimization algorithms with GPI-2

更多

 Asynchronous Stochastic Gradient Decent

 Direct RDMA access to main and GPU memory instead of message passing

 Optimized data-layers for distributed file

 High performance deep learning framework wiht our distributed runtime system

收起