NVIDIA A100 Tensor Core GPU datasheet (SXM4 and PCIe)

First-party source that resolves nvlink-3, which was previously sourced only to Wikipedia. The 'Interconnect' spec row reads 'NVIDIA NVLink Bridge for 2 GPUs: 600GB/s' and 'NVLink: 600GB/s' with footnote '** SXM4 GPUs via HGX A100 server boards; PCIe GPUs via NVLink Bridge for up to two GPUs'. Prose: 'NVIDIA NVLink in A100 delivers 2X higher throughput compared to the previous generation. When combined with NVIDIA NVSwitch, up to 16 A100 GPUs can be interconnected at up to 600 gigabytes per second (GB/sec)'. This is the authoritative first-party confirmation that the A100's 600 GB/s is the per-GPU BIDIRECTIONAL AGGREGATE across 12 links, not a per-direction rate.