Enable unprecedented levels of automation and agility with cloud computing solutions.
Deep Learning Solutions
Infrastructure for neural networks
Deep learning, a subset of artificial intelligence (AI), is about enabling computers to learn new concepts from raw data – much like the human mind. A deep learning system, for example, can spot the difference between a flower and a tree after viewing thousands of images of each. Deep learning requires extreme compute, I/O, and networks as well as exponential scaling. Traditional IT infrastructure is inadequate. To succeed with deep learning, users need optimal platforms and expertise.
The need for efficiency in deep learning
Deep learning has to be efficient to accomplish its business goals: Not only keeping operating costs down across the platform but also ensuring that users can quickly begin to apply deep learning applications. Hardware resources need to be sufficient to synthesise large volumes of data to drive insightful actions.
A purpose-built deep learning portfolio
HPE offers a comprehensive set of innovations to accelerate deep learning across all organisations. These innovations span systems design, partner ecosystem collaboration, and expertise including flexible consumption models from HPE Pointnext services. They complement the existing HPE Apollo portfolio and strengthen overall HPE capabilities from the Core Datacentre to the Intelligent Edge.
Performance from the most powerful GPUs
HPE deep learning solutions include the most powerful GPUs in a high GPU-to-CPU configuration, with high I/O bandwidth per node.
Scalability up to petaflop environment
Our new deep learning systems scale up to thousands of nodes, offering a range of environment sizes and flexible cost/performance characteristics.
Energy efficiency and productivity
We design for power efficiency and productivity, capping power at rack and node levels and offering fast time to solutions with powerful system tools.
Investing in GPUs for petaflop scale
We have invested in deep learning platforms ranging from entry-level GPU PCIe-enabled servers to platforms with the latest SXM2 NVLink GPUs.
Scale with unlimited infrastructure capacity
Add capacity in minutes, not months. Avoid long procurement cycles and save the costs of overprovisioning by managing capacity jointly with your account team.
Pay only for what you use
Align costs with usage monthly via advanced metering, with no upfront payment requirements. This approach fits analyst predictions that by 2020, 80% of IT infrastructure hardware and software purchases will be provided on a pay-as-you-go basis.
Combine hardware, software, and services costs
Pay a single price for hardware, software, and services. This applies to servers, software-defined and traditional storage, networks, converged, and software as well as HPE and multi-vendor support.
“Through our partnership with SGI, and now HPE, the Tokyo Institute of Technology has worked successfully to deliver a converged world-leading HPC and deep learning platform that can address our requirements and those of our nation. The NVIDIA Tesla P100 SXM2 node solution enables GPU based deep learning capability to be scalable to the entire size of our TSUBAME 3.0 system.”
HPE partner ecosystem for deep learning
We have developed an ecosystem to build out innovative solutions and offerings that address our customers’ deep learning needs.
Our solutions enable you to solve business problems with deep learning, e.g. the HPE Fraud Detection Solution with Kinetica.
We provide roadmap Integration for next-gen systems, engineering collaboration for new products, and testing for interoperability.
Technology adoption offerings
Our Centres of Excellence help you plan and deploy deep learning solutions and modernise your code.
A portfolio of deep learning Solutions
An ideal HPC and deep earning platform that provides unprecedented performance with 8 GPUs, high bandwidth fabric, a configurable GPU topology to match deep learning workloads, and a choice of high-speed, low latency fabrics with 2x IO expansion.
A liquid cooled, tray-based, scalable, high-density clustered computer system designed to deliver the utmost in performance, density, scale, and efficiency in an easy-to-manage, production-ready platform.
A 1U rackmount dual socket Intel Xeon Server with support for four NVIDIA Tesla SXM2 GPUs with NVLink. This configuration allows for up to 14,336 NVIDIA CUDA cores and 42.4 Tflops single precision floating point performance.
A cost effective 1U dual socket Intel Xeon server that’s optimal for AI/DL/ML workloads or a mixed-load HPC environment. With support for up to 4 PCIe GPU cards, Apollo pc40 provides up to 48 Tflops of single precision performance per server.
A solution for deep learning at the “Intelligent Edge” which performs real-time analysis of live data sets without having to engage the core datacentre.
HPE deep learning services
Be on-time, on-budget, and generally flawless with your deep learning implementation with our Professional and Operational Services. Take advantage of creative configurations to get the most out of your deep learning software and hardware.
Consume your deep learning infrastructure using a flexible, on-demand consumption model. Get scalable capacity as needed, paying only for what you use, including servers, storage, networks, software, and services.
Get early access
Join us for the HPE Early Access programme for 8-way SXM2 GPU Volta. This includes configuration, benchmarking, and technology selection guidance.
Learn how to manage object storage solutions with new breakthrough capabilities that are simple, cost effective, scalable, and can be rapidly deployed.
For engineering teams that demand CAE simulation turnaround in hours – not days or weeks – high-performance computing and ANSYS software can accelerate your breakthrough product design. Learn how to reduce design cycle time by 40 percent* when supporting multiple, concurrent complex simulations for prototyping. Register for a high-performance computing best practices guide for CAE. Explore how HPE Apollo 2000 systems offer higher density and scalability than competitors, with independent serviceability of networking modules enabling greater engineer productivity and competitive advantage for your business. Learn how data centre solutions like high-performance computing, together with ANSYS software, can speed computer-aided simulation and time-to-market while lowering TCO.
Digital transformation trends in IT services
What is deep learning?