NVIDIA BlueField Data Processing Unit
ARCHIVED 2023-11-22, DATE APPROXIMATE · VERSION 20231122_rev01 · COMPARED WITH 20231019_rev01
Full text changes — 20231019_rev01 to 20231122_rev01
COLOUR MARKS THE SEVERITY OF A FLAGGED CLAUSE · + AND − MARK ADDED AND REMOVED
| 1 | ## Transform the Data Center With NVIDIA DPUs | |
| 1 | ## Transform the Data Center With NVIDIA BlueField | |
| 2 | 2 | |
| 3 | NVIDIA BlueField® data processing units (DPUs) ignite unprecedented innovation for modern data centers and supercomputing clusters. By offloading, accelerating, and isolating a broad range of advanced networking, storage, and security services, BlueField DPUs provide a secure and accelerated infrastructure for any workload, in any environment, from cloud to data center to edge. | |
| 3 | The NVIDIA® BlueField® networking platform ignites unprecedented innovation for modern data centers and supercomputing clusters. With its robust compute power and integrated software-defined hardware accelerators for networking, storage, and security, BlueField creates a secure and accelerated infrastructure for any workload in any environment, ushering in a new era of accelerated computing and AI. | |
| 4 | 4 | |
| 5 | 5 | ### NVIDIA BlueField-3 DPU |
| 6 | 6 | |
| 7 | The NVIDIA BlueField-3 DPU is a 400 Gb/s infrastructure compute platform with line-rate processing of software-defined networking, storage, and cybersecurity. BlueField-3 combines powerful computing, high-speed networking, and extensive programmability to deliver software-defined, hardware-accelerated solutions for the most demanding workloads. From accelerated AI to hybrid cloud, high-performance computing to 5G wireless networks, BlueField-3 redefines the art of the possible. | |
| 7 | The NVIDIA BlueField-3 DPU is a 400 gigabits per second (Gb/s) infrastructure compute platform with line-rate processing of software-defined networking, storage, and cybersecurity. BlueField-3 combines powerful computing, high-speed networking, and extensive programmability to deliver software-defined, hardware-accelerated solutions for the most demanding workloads. From accelerated AI to hybrid cloud, high-performance computing to 5G wireless networks, BlueField-3 redefines the art of the possible. | |
| 8 | 8 | |
| 9 | ### NVIDIA BlueField-3 SuperNIC | |
| 10 | ||
| 11 | The BlueField-3 SuperNIC is a novel class of network accelerators based on the BlueField-3 networking platform, purpose-built for supercharging hyperscale AI workloads. Designed for network-intensive, massively parallel computing, the BlueField-3 SuperNIC provides up to 400Gb/s of remote direct-memory access (RDMA) over Converged Ethernet (RoCE) network connectivity between GPU servers, optimizing peak AI workload efficiency. Creating a new era of AI cloud computing, the BlueField-3 SuperNIC enables secure, multi-tenant data center environments with deterministic and isolated performance between jobs and tenants. | |
| 12 | ||
| 9 | 13 | ### NVIDIA BlueField-2 DPU |
| 10 | 14 | |
| 11 | 15 | The NVIDIA BlueField-2 DPU provides innovative acceleration, security, and efficiency in every host. BlueField-2 data center infrastructure combines the power of the NVIDIA ConnectX®-6 Dx with programmable Arm® cores and hardware offloads for software-defined storage, networking, security, and management workloads. |
| 12 | 16 | |
| 13 | 17 | NVIDIA BlueField-2 also delivers superior performance, security, and reduced total cost of ownership for cloud computing platforms, enabling organizations to efficiently build and operate virtualized, containerized, and bare-metal infrastructures at massive scale. |
| 14 | 18 | |
| 15 | ### NVIDIA Converged Accelerators | |
| 16 | ||
| 17 | NVIDIA converged accelerators combine the power of the NVIDIA® Ampere GPU architecture with the enhanced security and networking capabilities of the NVIDIA BlueField DPU, all in a single high-performance package. This advanced architecture delivers unprecedented performance and strong security for AI-powered workloads in edge computing, telecommunications, and network security. | |
| 18 | ||
| 19 | 19 | ## **The Conference for the Era of AI and the Metaverse** |
| 20 | 20 | |
| 21 | 21 | ### Developer Conference March 20-23 | Keynote March 21 |
| 22 | 22 | |
| 23 | 23 | Don't miss these three upcoming networking sessions at GTC. |
| 24 | 24 | |
| 25 | 25 | Join industry security experts to learn about the evolution of network security and the need for a zero-trust approach. Hybrid work demands enterprises to reevaluate and deliver security innovations for our workplace, workforce, and workload regardless of where workers and apps reside. As security workloads transition from centralized appliances at the edge to virtualized instances running on every server in the data center and cloud, powerful programmable processors are becoming essential. With the latest trends in AI, machine learning, 5G, and the internet of things, data center architects are being challenged to meet the increasing demands for scale and accelerated performance, while maintaining and innovating security at the highest level, despite an evolving threat landscape. We'll discuss the challenges of existing approaches and share techniques and strategies for a new way to offload, accelerate, and isolate security. |
| 26 | 26 | |
| 27 | 27 | In a world of growing infrastructure complexity and application sprawl, ensuring high performance, full stack security and operational agility has become a challenge for organizations that are creating competitive advantage. In this panel session, VMware and NVIDIA discuss essential insights into how businesses can leverage VMware Cloud Foundation and NVIDIA DPU and GPU technologies to extend infrastructure services through VMware vSphere on DPU. |
| 28 | 28 | |
| 29 | 29 | Computing is advancing at incredible speed, and cloud computing is becoming the norm for nearly every enterprise. Cloud computing allows companies to deploy applications faster, reduce operating costs, and rapidly scale resources to meet growing demand. However, not all clouds are created equal. Modern applications increasingly rely on accelerated computing as the underlying technology to deliver on the promise of machine learning and artificial intelligence. We'll discuss how organizations can build cloud computing infrastructures that are accelerated, efficient, and secure to power the next wave of applications. |
| 30 | 30 | |
| 31 | ### Data Center Networks Usher in the Era of AI | |
| 32 | ||
| 33 | The data center's network is ultimately responsible for ushering in the era of AI advancements and performance. Learn how the newest class of data centers can support AI workloads with accelerated computing and high-performance networking. | |
| 34 | ||
| 35 | ### The First Ethernet Network Designed for AI Workloads | |
| 36 | ||
| 37 | The NVIDIA Spectrum™-X networking platform, consisting of the NVIDIA Spectrum switch and the BlueField-3 SuperNIC, is the world's first Ethernet fabric built for AI, accelerating generative AI network performance by 1.6X over traditional Ethernet fabrics. | |
| 38 | ||
| 39 | ### Next-Generation Networking for the Next Wave of AI | |
| 40 | ||
| 41 | Modern AI workloads operate at data center-scale, heavily relying on fast and efficient connectivity between GPU servers. Discover the pivotal role of the BlueField-3 SuperNIC in enabling an optimal AI network. | |
| 42 | ||
| 31 | 43 | ### NVIDIA BlueField DPUs Drive Data Center Efficiency |
| 32 | 44 | |
| 33 | 45 | Learn how DPUs reduced power consumption in testing with key NVIDIA partners. |
| 34 | 46 | |
| 35 | 47 | ### DPU Power-Efficiency Research |
| 36 | 48 | |
| 37 | 49 | NVIDIA DPUs can reduce power consumption by 30%, saving $56 million for large data centers. Find all of NVIDIA's test results on the power consumption of various DPUs. |
| 38 | 50 | |
| 39 | 51 | ### Creating a Power-Efficient Data Center With DPUs |
| 40 | 52 | |
| 41 | 53 | Learn about BlueField DPU power savings research and get answers to data center efficiency questions. |
| 42 | 54 | |
| 43 | ### Experience VMware vSphere on NVIDIA BlueField DPUs | |
| 55 | BlueField accelerates AI performance with robust RoCE networking, GPUDirect® RDMA and RoCE, direct data placement, and programmable congestion control, creating an exceptionally fast and efficient network infrastructure for AI. | |
| 44 | 56 | |
| 45 | VMware vSphere is available on NVIDIA LaunchPad, providing easy access to VMware ESXi-optimized software running on BlueField DPU-accelerated infrastructure for testing next-generation applications and workloads. | |
| 57 | BlueField enables a zero-trust, security-everywhere architecture where security goes beyond the data center perimeter to the edge of every server. | |
| 46 | 58 | |
| 47 | ### Test and Run Red Hat OpenShift on NVIDIA BlueField DPUs | |
| 59 | With NVMe over Fabrics (NVMe-oF), GPUDirect Storage, encryption, elastic storage, data integrity, decompression, and deduplication, BlueField provides high-performance storage access with latencies for remote storage that rival direct-attached storage. | |
| 48 | 60 | |
| 49 | OpenShift and BlueField provide a consistent, cloud-native application platform for managing hybrid-cloud, multi-cloud, and edge deployments with enhanced orchestration, automation, and security. Apply to test and run this solution. | |
| 61 | BlueField is a powerful cloud infrastructure processor, delivering up to 400Gb/s of Ethernet and InfiniBand connectivity for both traditional applications and modern workloads, while freeing host CPU cores to run applications instead of infrastructure tasks. | |
| 50 | 62 | |
| 51 | BlueField DPUs enable a zero-trust, security-everywhere architecture where security goes beyond the data center perimeter to the edge of every server. | |
| 63 | ### Experience VMware vSphere on NVIDIA BlueField DPUs | |
| 52 | 64 | |
| 53 | With NVMe over Fabrics (NVMe-oF), GPUDirect Storage, encryption, elastic storage, data integrity, decompression, and deduplication, BlueField provides high performance storage access with latencies for remote storage that rival direct attached storage. | |
| 65 | VMware vSphere is available on NVIDIA LaunchPad, providing easy access to VMware ESXi-optimized software running on BlueField DPU-accelerated infrastructure for testing next-generation applications and workloads. | |
| 54 | 66 | |
| 55 | BlueField is a powerful data center services accelerator, delivering up to 400 gigabits per second (Gb/s) of Ethernet and InfiniBand connectivity for both traditional applications and modern GPU-accelerated workloads, while freeing host CPU cores to run applications instead of infrastructure tasks. | |
| 67 | ### BlueField-Enhanced Servers Supercharge Generative AI for Enterprises | |
| 56 | 68 | |
| 57 | The NVIDIA DOCA SDK enables developers to easily create high-performance, software-defined, cloud-native, DPU-accelerated services, leveraging industry-standard APIs. | |
| 69 | The world's leading system manufacturers are building AI-ready servers, featuring NVIDIA L40S GPUs, NVIDIA BlueField-3 DPUs and SuperNICs, and NVIDIA AI Enterprise software, to help enterprises customize and deploy generative AI applications. | |
| 58 | 70 | |
| 59 | ### Transform HPC Data Centers With NVIDIA BlueField DPUs | |
| 71 | ### Transform HPC Data Centers With NVIDIA BlueField | |
| 60 | 72 | |
| 61 | Providing powerful computing, high-speed InfiniBand networking, and highly programmable engines, the NVIDIA BlueField DPU ignites unprecedented innovation and performance for modern scientific computing applications. | |
| 73 | Providing powerful computing, high-speed InfiniBand networking, and highly programmable engines, the NVIDIA BlueField networking platform ignites unprecedented innovation and performance for modern scientific computing applications. | |
| 62 | 74 | |
| 63 | ### Take Supercomputing Performance Into the Cloud | |
| 64 | ||
| 65 | The NVIDIA Cloud-Native Supercomputing platform pairs the NVIDIA BlueField DPU architecture with high-speed, low-latency NVIDIA Quantum InfiniBand networking to deliver bare-metal performance, user management and isolation, data protection, and on-demand HPC and AI services-simply and securely. | |
| 66 | ||
| 67 | 75 | ### Boost Cybersecurity With the NVIDIA Morpheus AI Framework |
| 68 | 76 | |
| 69 | 77 | NVIDIA Morpheus is an AI application framework that provides cybersecurity developers with a highly optimized AI pipeline and pretrained AI capabilities that bring a new level of security to data centers. The Morpheus framework provides highly scalable and adaptive data ingest, pretrained AI models, and policy control functions, enabling cybersecurity developers to detect and remediate security threats. |
| 70 | 78 | |
| 71 | ### Deliver Metaverse Applications at Scale With NVIDIA OVX Systems | |
| 79 | ### Deliver High-Performance, Software-Defined 5G With NVIDIA AX800 | |
| 72 | 80 | |
| 73 | Learn how NVIDIA BlueField DPUs accelerate the creation and operation of [NVIDIA Omniverse™](https://www.nvidia.com/en-eu/omniverse/) applications running on NVIDIA OVX™ systems. | |
| 81 | The NVIDIA AX800 converged accelerator delivers high-performance 5G, with deep integration of the BlueField-3 DPU and A800 GPU, allowing cloud service providers to host 5G RAN and AI services on multi-tenant GPU cloud infrastructure. | |
| 74 | 82 | |
| 75 | ### Accelerate Ethernet for AI with NVIDIA Spectrum-X Networking Platform | |
| 83 | ### Oracle Cloud Infrastructure Chooses NVIDIA BlueField Networking | |
| 76 | 84 | |
| 77 | NVIDIA Spectrum™\-X networking platform, featuring Spectrum-4 switches and BlueField-3 DPUs, is the world's first Ethernet fabric built for AI, accelerating generative AI performance by 1.7x over traditional Ethernet fabrics. | |
| 85 | Oracle Cloud Infrastructure (OCI) is advancing sustainability goals by using BlueField DPUs to offload data center tasks from CPUs, achieving greater performance, efficiency, and security. | |
| 78 | 86 | |
| 79 | ### Deliver High-Performance, Software-Defined 5G with NVIDIA AX800 | |
| 87 | - Systems | |
| 88 | - Cloud | |
| 89 | - Infrastructure | |
| 90 | - Security | |
| 91 | - Storage | |
| 80 | 92 | |
| 81 | The NVIDIA AX800 converged accelerator delivers high-performance 5G, with deep integration of the BlueField-3 DPU and A800 GPU, allowing cloud service providers to host 5G RAN and AI services on multi-tenant GPU cloud infrastructure. | |
| 82 | ||
| 83 | - **Systems** | |
| 84 | - **Infrastructure** | |
| 85 | - **Security** | |
| 86 | - **Storage** | |
| 87 | ||
| 88 | 93 | ### Try the DOCA SDK |
| 89 | 94 | |
| 90 | 95 | Sign up to receive exclusive news, updates, and early access to the NVIDIA DOCA SDK. |