Nvidia recently introduced its Spectrum-X network platform. This platform should provide high-speed Ethernet connectivity for AI workloads across multiple cloud environments.
platform will According to Nvidia Advance the development of multi-tenant AI cloud environments. This is by improving the efficiency of the necessary Ethernet networks and how data is transmitted.
Ultimately, Nvidia wants to make it easier for cloud providers, who use an Ethernet connection instead of an Infiniband, to train and run AI models in the cloud.
Platform configuration
The Spectrum-X platform consists of Nvidia’s proprietary Ethernet switching technology, sourced from Mellanox, in combination with a BlueField-3 DPU. This environment can accelerate the performance and energy efficiency of AI workloads in a stack by a factor of 1.7.
Spectrum-4 switches can be configured into 64800 GbE ports or 128400 GbE ports. It also facilitates a throughput of 51.2 terabits per second for switching and routing.
attractive to cloud service providers
In addition, Nvidia’s Spectrum-X is interoperable with existing Ethernet packages, making it more attractive for cloud providers to run AI workloads on existing cloud infrastructure.
Concretely, data centers are supported for cloud environments to process data and deliver more efficient AI workloads, without compromising the already existing Ethernet-based multi-tenant model.
Major infrastructure providers such as Dell Technologies, Lenovo and Supermicro Nvidia Spectrum-X have also embraced it.
Israeli supercomputer
Nvidia is currently developing a supercomputer in Israel that uses Spectrum-X. This Israel-1 supercomputer will become a test environment for Spectrum-X. The supercomputer consists of Dell PowerEdge XE9680 servers with eight Nvidia H100 GPUs. This should result in eight exaflops of AI performance and 130 petaflops for scientific computing loads.
Read also: Nvidia will take first prize for artificial intelligence chips
