CRAY ACCEL AI DEEP LEARNING
Everything you need to go from pilot to production
Artificial intelligence (AI) and, in particular, deep learning (DL) are rapidly transforming entire industries and scientific disciplines. If your toe isn’t already in the AI/DL water, it might feel like you’re already too late. You’re not — because Cray can help. How? By giving you the tools and support to guide you to success whether you’re just starting out or ready for production AI.
Our Cray Accel AI offerings leverage our supercomputing expertise, technologies and best practices into solutions that advance the adoption of deep learning. These fast-start configurations range from a starter system ideal for AI exploration to a complete, production-level cluster supercomputer for training and inference.
At Cray, we have a long track record supporting AI/DL and other complex computing challenges. The high-performance techniques that make AI possible today have been honed over decades in other technologies such as medical imaging, cybersecurity, climate modeling and seismic processing — technologies driven to success by our supercomputing systems. So wherever you are on your AI journey, Cray has the high-powered computing systems and the industry expertise to make AI and DL work for you.
Test, launch and grow your deep learning initiatives
The Cray Accel AI solutions leverage the enterprise-ready, industry-standards-based, fully scalable Cray® CS-Storm™ accelerated cluster supercomputer featuring NVIDIA® Tesla® V100 GPU accelerators powered by the NVIDIA® Volta™ GPU architecture. Every Accel AI solution comes with a robust deep learning environment from Bright Computing that includes TensorFlow™, MXNet, Caffe2, Chainer, Microsoft Cognitive Toolkit and more.
Choose from three fast-start configurations: the Accel AI Quick Start for initial deep learning trials, the Accel AI Cluster Starter Kit for deep learning exploration and small proof-of-concept projects, or the Accel AI Deep Learning System for production-level deep learning training and inference.
Accel AI Quick Start
If you're at the tool exploration and model development stage, the Accel AI Quick Start solution delivers all the elements necessary for a small team to get started with deep learning trials. You get a single-chassis, fully configured Cray CS-Storm 500NX server with NVIDIA Tesla Volta GPUs using NVLink and running Bright Computing's comprehensive deep learning environment.
Learn more about a special limited time offer
Accel AI Cluster Starter Kit
The Accel AI Cluster Starter Kit offers the technology you need for application development, initial production and proof-of-concept projects. You get two CS-Storm 500NX servers with the same NVIDIA components and Bright Computing software environment as the Quick Start solution but add a management node and networking, allowing you to manage the system and connect it directly to your existing cluster infrastructure.
Accel AI Deep Learning System
When you're ready for a production-level system, choose the Accel AI Deep Learning System. This solution gives you four CS-Storm 500NX servers with the same NVIDIA components as the Quick Start and Starter Kit solutions, then adds a complete management and networking infrastructure with high-speed Ethernet® and InfiniBand® as well as shared storage.
Complete deep learning software environment
The comprehensive deep learning software environment from Bright Computing provides the AI frameworks, libraries and tools you need for complex machine and deep learning workloads.
Machine and deep learning frameworks include Cafe, Caffe2, TensorFlow, MXnet, Microsoft Cognitive Toolkit, Torch, Theano
Machine learning libraries include MLPython, NVIDIA CUDA Deep Neural Network library (cuDNN) and Deep Learning GPU Training System
Supporting infrastructure includes over 400 MB of Python modules that support the machine learning packages, plus the NVIDIA hardware drivers, CUDA (parallel computing platform API) drivers, CUB (CUDA building blocks) and NCCL (library of standard collective communication routines)