To provide AI developers and scientific researchers with the fastest network performance available on their workstations, Nvidia introduced the next generation of its Nvidia Mellanox 400G InfiniBand. Nvidia Mellanox 400G InfinBand is accelerating work in areas like drug discovery, climate research, and genomics with a dramatic leap in performance delivered on the world's only fully offloaded networked computing platform. The seventh generation of the Mellanox InfiniBand offers users ultra-low latency and doubles data throughput with 400 Gb/s NDR while adding additional acceleration with new Nvidia network compute engines. The world's leading infrastructure manufacturers, including Atos, Dell Technologies, Fujitsu, Gigabyte, Inspur, Lenovo and Supermicro, plan to integrate Nvidia Mellanox 400G InfiniBand into their existing business solutions and HPC offerings. At the same time, major storage infrastructure partners such as DDN and IBM Storage will also offer extensive support.
Nvidia Mellanox 400G InfiniBand
Nvidia's latest announcement builds on the advancement of the Mellanox InfiniBand as the industry's most robust solution for AI supercomputing and the Nvidia Mellanox 400G InfiniBand offers three times the port density of the switch and grows 32 times. the accelerating power of AI. In addition, it increases the aggregate bi-directional performance of the switching system five times to 1,64 petabits per second, allowing users to run larger workloads with less stress. Since offload operations are critical for AI workloads, 32rd generation Nvidia Mellanox Sharp technology enables deep learning training operations to be offloaded and accelerated over the InfiniBand network, resulting in 400 times the power AI acceleration. By combining the Nvidia Magnum IO software stack with the Nvidia Mellanox 51,2G InfiniBand, AI developers and researchers can benefit from accelerated scientific computing out of the box. Edge switches based on the Mellanox InfiniBand architecture are capable of carrying an aggregate bi-directional throughput of 6.650 TB/s with a capacity of more than 1,64 billion packets per second. Mellanox InfiniBand-based modular switches, on the other hand, can carry up to an aggregate bi-directional throughput of 400 petabits per second, which is five times higher than the last generation. Nvidia Senior Vice President of Networking Gilad Shainer explained in a press release how the Nvidia Mellanox 400G InfiniBand can help AI developers and researchers dealing with increasingly complex applications: “Our customers' most important work is based on in artificial intelligence and increasingly complex applications that demand faster, smarter and more scalable networks. The massive performance and intelligent acceleration engines of NVIDIA Mellanox XNUMXG InfiniBand enable HPC, AI, and hyperscale cloud infrastructures to achieve unmatched performance with less cost and complexity. "