DDN and NVIDIA support the development of artificial intelligence
Artificial intelligence solutions widely available for data-centric businesses
DDN's A3I Series solutions supporting NVIDIA DGX A100 servers facilitate the deployment of innovative solutions using artificial intelligence, offering high performance, seamless scalability and simplified operations.
Artificial Intelligence (AI) and Deep Learning (DL) are transforming every industry in the world. They are creating the most aggravating data processing tasks in the history of computer processing and are posing major design challenges for storage, data processing and network infrastructure, including architecture using A100 devices, so they can be very complex and take a long time to implement.
DDN A3I solutions to quickly and easily deploy an optimized infrastructure for artificial intelligence fully integrate DDN's storage appliances and NVIDIA DGXTM A100 series servers. They provide superior application performance and scale seamlessly to meet the demands of increasingly system-intensive artificial intelligence workloads.
DDN A3I with NVIDIA DGX A100: high-speed processing, aided by artificial intelligence
DDN A3I (Accelerated, Any-Scale AI) solutions cross barriers and provide a new foundation for artificial intelligence and deep learning. Built from the ground up for data centers designed for artificial intelligence, DDN A3I solutions together with NVIDIA DGX A100 servers accelerate the entire data processing process for any-scale artificial intelligence and deep learning processes. They are designed to deliver extreme performance and capacity from a jointly designed and proven architecture. DDN A3I along with NVIDIA DGX A100 are optimized at every layer of hardware and software, providing fast data flow and reliable and consistent storage access.
The shared, parallel DDN architecture allows full utilization of the DGX A100's graphics processing unit (GPU) resources and maximizes developer productivity in each DGX A100 server processing cycle. A3I's DDN solutions support deep learning, enabling simultaneous continuous processing of all phases of tasks, including data acquisition, selection, training, checking, inference, and simulation.
With DDN's A3I solutions, data can be accessed simultaneously by multiple DGX A100 servers through a fully optimized, easy-to-use unified interface by containerized applications.
Easy to deploy, manage and use artificial intelligence infrastructure
DDN A3I, together with NVIDIA DGX A100, offers a turnkey, pre-configured solution that is easy to deploy, reducing the required time from creating artificial intelligence assumptions to delivering business results in a production environment. Based on DDN A3I's comprehensive reference architecture, these solutions eliminate the work involved in design selection. In collaboration with NVIDIA, they have been validated to provide the highest performance, optimal efficiency and the ability to grow flexibly with DGX A100 servers.
This powerful solution, based on NVIDIA DGX A100 artificial intelligence supercomputers as its processing foundation, delivers performance of more than 1 petaFLOPS of deep learning operations. It uses eight NVIDIA Tesla V100 Tensor Core GPUs configured in a technology called hybrid cube-mesh (each with each) with a GPU communication network built on NVIDIA NVLink to guarantee ultra-high throughput and minimal latency. The DGX A100 architecture overcomes the performance bottlenecks of traditional GPU interconnects and offers predictable linear performance across multiple GPUs.
The DGX A100 uses NVIDIA DGX software that is optimized in every layer, including the most popular deep learning platforms along with the necessary drivers and libraries. As a result, DDN A3I together with NVIDIA DGX A100 guarantees to achieve unparalleled performance of multi-system artificial intelligence applications using multiple GPUs. To meet the expectations of a variety of demanding DDN A3I tasks with NVIDIA, DGX A100 uses DDN AI200X, AI400X and AI7990X storage devices.
The AI200X and AI400X are fully integrated parallel file system storage devices using only NVME drives, achieving application throughput of up to 48GB/s and more than 3 million IOPS, accelerating even the most demanding tasks. The AI200X and AI400X are specially optimized to take full advantage of the GPU computing capabilities of the DGX A100, providing maximum efficiency while making demanding data processing operations easy to manage.
The AI7990X is a hybrid storage appliance with a parallel file system, integrating flash drives and high-capacity traditional disks in a unified system for simplicity and flexibility. This integration easily allows for simple colocation of both hot training data and large libraries, while ensuring optimal system efficiency. The AI7990X outperforms competing platforms and cost-effectively delivers large capacity, for growing data libraries. DDN storage devices, working with NVIDIA DGX A100, are fully integrated and optimized for demanding artificial intelligence and deep learning tasks.
The DDN architecture allows for instant provisioning of new resources for applications and easy deployment of additional NVIDIA DGX A100 servers. Advanced monitoring tools, built into all components of the solution, provide easy data management capabilities and comprehensive metrics for optimizing running online tasks. A robust digital security infrastructure makes it easy to deploy a secure platform with multiple independent environments (multitenancy) and share the power of the DGX A100 server with a wide range of users, while ensuring data access and regulatory compliance.
AI200X™
AI400X™
AI7990X™
