We use cookies 🍪
We use cookies and other similar technologies to improve your browsing experience and the functionality of our site. Learn more in our Privacy Policy.

NVIDIA A2 16GB PCI-e 4.0 x8 GDDR6 CUDA Graphics Accelerator Card - Used

NVIDIA A2 16GB PCI-e 4.0 x8 GDDR6 CUDA Graphics Accelerator Card - Used
NVIDIA A2 16GB PCI-e 4.0 x8 GDDR6 CUDA Graphics Accelerator Card - Used
NVIDIA A2 16GB PCI-e 4.0 x8 GDDR6 CUDA Graphics Accelerator Card - Used
NVIDIA A2 16GB PCI-e 4.0 x8 GDDR6 CUDA Graphics Accelerator Card - Used
NVIDIA A2 16GB PCI-e 4.0 x8 GDDR6 CUDA Graphics Accelerator Card - Used
NVIDIA A2 16GB PCI-e 4.0 x8 GDDR6 CUDA Graphics Accelerator Card - Used
NVIDIA A2 16GB PCI-e 4.0 x8 GDDR6 CUDA Graphics Accelerator Card - Used
$564.99
  • Condition: Used
  • Stock: 10
  • Brand: NVIDIA
  • Model: NVIDIA A2 16GB
  • Weight: 5.00lb
  • MPN: NVIDIA A2 16GB
Product Description

The NVIDIA A2 16GB PCIe 4.0 x8 GDDR6 CUDA Graphics Accelerator Card is a compact, low-power data center GPU designed for AI inference, edge computing, intelligent video analytics, and accelerated workloads. Built on NVIDIA Ampere architecture, the A2 features 16GB of GDDR6 memory with ECC support and 200GB/s memory bandwidth. It provides 4.5 TFLOPS of peak FP32 performance, along with Tensor Core acceleration for AI inference tasks. The card uses a PCI Express 4.0 x8 interface and a single-slot, low-profile design, making it suitable for space- constrained server deployments. With configurable 40–60W power consumption and a passive cooling solution, the NVIDIA A2 is designed for efficient operation in compatible server environments with adequate chassis airflow.

Key Specifications
Manufacturer
NVIDIA
Product Family
NVIDIA A-Series
Model
NVIDIA A2 16GB
Product Type
Tensor Core GPU Accelerator Card
Architecture
NVIDIA Ampere
GPU Chip
GA107
CUDA Cores
1,280
RT Cores
10
Compute Capability
8.6
Peak FP32 Performance
4.5 TFLOPS
TF32 Tensor Core Performance
9 TFLOPS / 18 TFLOPS with Sparsity
Peak FP16 Tensor Core Performance
18 TFLOPS / 36 TFLOPS with Sparsity
GPU Memory
16GB
Memory Type
GDDR6 with ECC
Memory Bus Width
128-bit
Memory Bandwidth
200 GB/s
Graphics Interface
PCI Express 4.0 x8
Form Factor
Low-Profile, Single-Slot PCIe
Cooling
Passive Cooling
Maximum Power Consumption
40–60W (Configurable)
Video Encoder
1x NVENC
Video Decoders
2x NVDEC (Includes AV1 Decode)
ECC Memory Support
Yes
Intended Use
AI Inference / Edge Computing / Video Analytics
Condition
Used
Features
✓16GB GDDR6 ECC Memory
✓NVIDIA Ampere Architecture
✓1,280 CUDA Cores
✓10 RT Cores
✓200 GB/s Memory Bandwidth
✓PCIe 4.0 x8 Interface
✓4.5 TFLOPS FP32 Performance
✓Hardware AV1 Video Decode
✓40–60W Configurable Power
✓Low-Profile, Single-Slot Design
Compatible With
✓PCIe 4.0 x8 Server Platforms
✓Compatible Low-Profile PCIe Server Slots

Important: The NVIDIA A2 is a passively cooled data center accelerator designed for server use. It requires adequate chassis airflow, a compatible PCIe slot, and suitable power delivery. Confirm server compatibility and physical clearance before installation.

Warranty & Returns

Covered by our standard hardware warranty. Tested before dispatch. Extended warranty available. Returns accepted within 30 days — see our Return Policy.

Shipping

Worldwide shipping available; tracking provided. In-stock items usually dispatch within 1 business day.

```