NVIDIA HGX A100 datasheet

Capacity-unstated A100 copy of [[sources/link2-nvidia-hgx-a100-datasheet]], added 2026-10-10 for [[accelerators/nvidia-a100-sxm]]. SAME DOCUMENT, SAME URL. First-party NVIDIA platform datasheet for the 4/8/16-GPU HGX A100 configurations, backing the capacity-unstated record's fabric: NVIDIA NVLink '3rd generation', NVIDIA NVSwitch 'N/A' at 4 GPUs and '2nd generation' at 8 and 16, GPU-to-GPU Bandwidth 600 GB/s at all three sizes. The scale-up fabric is identical across the two A100 SXM memory cuts, so this source is capacity-independent for the interconnect rows. All Tensor Core rows carry the '* With sparsity' footnote.