Cray and NVIDIA Supercomputing Solutions
Cray and NVIDIA work together to create innovative accelerated GPU supercomputer solutions for research, simulation and deep learning use cases. These solutions feature the latest NVIDIA® Tesla® GPU accelerators and cutting-edge hybrid system configurations. With a solution for every budget and requirement, Cray and NVIDIA make it easy to take advantage of the latest advances in GPU computing.
Cray® XC™ Series Supercomputers
The Cray® XC™ series supercomputer combines scalar, network and many-core innovation to create true, productive hybrid supercomputers to address virtually any type of high-performance application requirement.
Cray’s adaptive supercomputing architecture supports upgradability and best-in-class processing technologies, including the NVIDIA® Tesla® GPU family of accelerators. The Cray XC series supercomputer provides a unique Aries™ high-performance interconnect and flexible Dragonfly network topology, delivering low latency and scalable global bandwidth to satisfy the most challenging multi-petaflops applications.
Additionally, many-core accelerators lead to increased energy efficiency. Applications using the XC series deliver industry-leading energy efficiency when measured for real-world, sustained application workloads.
Cray’s differentiated software programming environment includes a mature compiler and performance enhancing analysis tool suite, and it supports numerous code development models including ISVs, CUDA and directives-based parallel programming methods.
Cray® CS-Storm™ Cluster Supercomputers
Cray’s CS-Storm™ accelerated GPU cluster is a system that consists of multiple high-density multi-GPU server nodes, designed for massively parallel computing GPU workloads like deep learning training. Each CS-Storm rack can hold up to 22 2U rackmount server nodes. Server nodes include two CPU processors, and from four to eight NVIDIA® Tesla® GPUs delivering up to 850 GPU TF of compute performance in one 48U rack.