Skip to content
NVNEXUS
GPU architecture

NVIDIA Ampere 2020 to 2022

NVIDIA Ampere brought third-generation Tensor Cores with TF32 and structural sparsity, Multi-Instance GPU and HBM2e memory to the data center with the A100, and cost-effective inference and virtual desktop cards with the A2, A10 and A16. The same architecture powers the Jetson Orin family and the RTX A-series workstation cards, so Ampere remains widely deployed at the edge and in the enterprise.

  • Third-generation Tensor Cores with TF32 and sparsity
  • Multi-Instance GPU (A100)
  • HBM2e on A100, GDDR6 on inference and workstation cards
  • Second-generation RT Cores
  • Shared with Jetson Orin edge modules

Showing all 5 GPUs

Also built on Ampere

Systems and modules that use Ampere GPUs.