Nvidia Quantum-2, Next Gen of IB Networking Platform
400Gb/s IB switch features secure, cloud-native, multi-tenant, bare-metal performance for AI, data analytics, and HPC applications.
This is a Press Release edited by StorageNewsletter.com on November 19, 2021 at 2:01 pmNvidia Corp. announced Quantum-2, the next gen of its IB networking platform, which offers performance, broad accessibility and security needed by cloud computing providers and supercomputing centers.
An advanced end-to-end networking platform, Quantum-2 is a 400Gb/s IB networking platform that consists of the Quantum-2 switch, the ConnectX-7 network adapter, the BlueField-3 data processing unit (DPU) and the software that supports the new architecture.
The introduction of Quantum-2 comes as supercomputing centers are increasingly opening to multitudes of users, many from outside their organizations. At the same time, the world’s cloud service providers are beginning to offer more supercomputing services to their millions of customers.
Quantum-2 includes features required for demanding workloads running in either arena. Supercharged by cloud-native technologies, it provides performance with 400Gb/s of throughput and multi-tenancy to accommodate many users.
“The requirements of today’s supercomputing centers and public clouds are converging,” said Gilad Shainer, SVP, networking, Nvidia. “They must provide the greatest performance possible for next-gen HPC, AI and data analytics challenges, while also securely isolating workloads and responding to varying demands of user traffic. This vision of the modern data center is now real with Nvidia Quantum-2 IB.”
Quantum-2 performance and cloud-native capabilities
With 400Gb/s, Quantum-2 IB doubles the network speed and triples the number of network ports. It accelerates performance by 3x and reduces the need for data center fabric switches by 6x, while cutting data center power consumption and reducing data center space by 7% each.
The multi-tenant performance isolation of Quantum-2 keeps the activity of one tenant from disturbing others, utilizing a telemetry-based congestion control system with cloud-native capabilities that ensure reliable throughput, regardless of spikes in users or workload demands.
Quantum-2 SHARPv3 In-Network Computing technology provides 32x more acceleration engines for AI applications compared with the previous gen. IB fabric management for data centers, including predictive maintenance, is enabled with the Nvidia UFM Cyber-AI platform.
A nanosecond-precision timing system integrated into Quantum-2 can synchronize distributed applications, like database processing, helping to reduce the overhead of wait and idle times. This capability allows cloud data centers to become part of the telecommunications network and host software-defined 5G radio services.
A100 GPU
Quantum-2 IB switch
At the heart of the Quantum-2 platform is the Quantum-2 IB switch. With 57 billion transistors on 7-nanometer silicon, it is slightly bigger than the A100 GPU with 54 billion transistors.
It features 64 ports at 400Gb/s or 128 ports at 200Gb/s and will be offered in a variety of switch systems up to 2,048 ports at 400Gb/s or 4,096 ports at 200Gb/s – more than 5x the switching capability over the previous gen, Quantum-1.
The combined networking speed, switching capability and scalability is for building the next-gen of giant HPC systems.
The Quantum-2 switch is available from a range of infrastructure and system vendors around the world, including Atos, DataDirect Networks, Dell Technologies, Excelero, Gigabyte, HPE, IBM, Inspur, Lenovo, NEC, Penguin Computing, QCT, Supermicro, VAST Data and WekaIO.
Infiniband Connectx 7
Quantum-2, ConnectX-7 and BlueField-3
The Quantum-2 platform provides 2 networking end-point options, the firm’s ConnectX-7 NIC and BlueField-3 DPU IB.
ConnectX-7, with eight billion transistors in a 7-nanometer design, doubles the data rate of the current leading HPC networking chip, the ConnectX-6. It also doubles the performance of RDMA, GPUDirect Storage, GPUDirect RDMA and In-Networking Computing. The ConnectX-7 will sample in January.
BlueField-3 IB, with 22 billion transistors in a 7-nanometer design, offers sixteen 64-bit Arm CPUs to offload and isolate the data center infrastructure stack. BlueField-3 samples in May.
Resources:
Video: Announcing NVIDIA Quantum-2 InfiniBand
Video: In-Network Computing with NVIDIA SHARP
Quantum-2 InfiniBand Platform