When you purchase through links on our site, we may earn an affiliate commission.Heres how it works.

The platform has been deployed at Colossus, the worlds largest AI supercomputer, since its inception.

Naturally, training these models requires significant web link performance.

Nvidia H100

xAI revealed it has been able to maintain 95% data throughput enabled by Spectrum-Xs congestion control capabilities.

The company added this level of performance cannot be delivered at this scale via standard Ethernet.

xAI opted to combine the Spectrum-X SN5600 switch with NVIDIA BlueField-3 SuperNICs for higher performance.

You might also like