NVIDIA A2 Tensor Core GPU

The NVIDIA® A2 Tensor Core GPU provides entry-level inference with low power, a small footprint, and high performance for NVIDIA AI at the edge. Featuring a low-profile PCIe Gen4 card and a low 40-60W configurable thermal design power (TDP) capability, the A2 brings versatile inference acceleration to any server for deployment at scale.


1 With sparsity
2 Supported in future vGPU release

Download NVIDIA A2 Tensor Core GPU datasheet

XENON NVIDIA A2
SPECIFICATIONS
NVIDIA A2 Tensor Core GPU
SPECIFICATIONS

Peak FP3

NVIDIA A2 Tensor Core GPU

4.5 TF

SPECIFICATIONS

TF32 Tensor Core

NVIDIA A2 Tensor Core GPU

9 TF | 18 TF¹

SPECIFICATIONS

BFLOAT16 Tensor Core

NVIDIA A2 Tensor Core GPU

18 TF | 36 TF¹

SPECIFICATIONS

Peak FP16 Tensor Core

NVIDIA A2 Tensor Core GPU

18 TF | 36 TF¹

SPECIFICATIONS

Peak INT8 Tensor Core

NVIDIA A2 Tensor Core GPU

36 TOPS | 72 TOPS¹

SPECIFICATIONS

Peak INT4 Tensor Core

NVIDIA A2 Tensor Core GPU

72 TOPS | 144 TOPS¹

SPECIFICATIONS

RT Cores

NVIDIA A2 Tensor Core GPU

10

SPECIFICATIONS

Media engines

NVIDIA A2 Tensor Core GPU

1 video encoder 2 video decoders (includes AV1 decode)

SPECIFICATIONS

GPU memory

NVIDIA A2 Tensor Core GPU

16GB GDDR6

SPECIFICATIONS

GPU memory bandwidth

NVIDIA A2 Tensor Core GPU

200GB/s

SPECIFICATIONS

Interconnect

NVIDIA A2 Tensor Core GPU

PCIe Gen4 x8

SPECIFICATIONS

Form factor

NVIDIA A2 Tensor Core GPU

1-slot, low-profile PCIe

SPECIFICATIONS

Max thermal design power (TDP)

NVIDIA A2 Tensor Core GPU

40–60W (configurable)

SPECIFICATIONS

Virtual GPU (vGPU) software support²

NVIDIA A2 Tensor Core GPU

NVIDIA Virtual PC (vPC), NVIDIA Virtual Applications (vApps), NVIDIA RTX Virtual Workstation (vWS), NVIDIA AI Enterprise, NVIDIA Virtual Compute Server (vCS)

Quick Quote Request

  • This field is for validation purposes and should be left unchanged.

Get a Quote Talk to a Solutions Architect