Architecture Ampere Chipset A100 Cuda cores 6912 Tensor cores 432 RT cores   Occupied slots 2 Bus PCIe 4.0 x16 memory 80GB HBM2e Connections none DirectX Not available OpenGL Not available Shader Model Not available Cooling type Passive Microsoft DDA support Yes VMware vDGA support No VMware vSGA support No Recommended use AI, Virtualization Software PyTorch, TensorFlow