NVIDIA A2 16GB
Overview
The NVIDIA A2 is the entry-level Ampere datacenter card, a passively cooled, half-height, half-length single-slot PCIe board built to add inference to servers that cannot take a full-size accelerator, from 5G edge boxes to existing CPU nodes. It carries 16 GB of GDDR6 on a 128-bit bus at 200 GB/s and connects over PCIe Gen4 x8, and its board power is configurable from 40 to 60 W. NVIDIA rates it at 4.5 TFLOPS of FP32, 9 TFLOPS of TF32, 18 TFLOPS of FP16 and BF16 and 36 TOPS of INT8 dense, each doubled with structured sparsity, plus 72 dense TOPS of INT4. It also has ten RT cores, one video encoder and two decoders with AV1 decode, so it doubles as a video analytics and virtual desktop card. NVIDIA introduced it at GTC in November 2021.
Performance
Peak theoretical throughput by precision type
| Precision | Dense | 2:4 Sparse |
|---|---|---|
FP64 | No verified data available | Structured sparsity is a tensor-core feature; this vector precision has no sparse form |
FP32 32-bit floating point | 4.5TFLOPS | Structured sparsity is a tensor-core feature; this vector precision has no sparse form |
TF32 TensorFloat-32 | 9TFLOPS | 18TFLOPS |
BF16 Brain Float 16 | 18TFLOPS | 36TFLOPS |
FP16 16-bit floating point | 18TFLOPS | 36TFLOPS |
FP8 | No verified data available | The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision |
FP6 | No verified data available | The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision |
FP4 | No verified data available | The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision |
INT8 8-bit integer | 36TOPS | 72TOPS |
INT4 4-bit integer | 72TOPS | 144TOPS |
Dense peak divided by accelerator TDP. Board power only: excludes host CPUs, networking, cooling and facility overhead. TDP for this part: 60 W.
The A2 16GB in the GPU landscape
Peak FP16 TFLOPS (dense) against TDP, single-GPU parts tracked by Flopper
Higher and further left is better: more half-precision throughput for less power.
Specifications
Architecture
Ampere
Form Factor
PCIe
Launch Year
2021
Process Node
No verified data available
Memory
16 GB GDDR6
Bandwidth
200 GB/s
TDP
60 W
Max power
60 W
Spec Confidence
Official
Full Specifications
| Compute Engine | |
|---|---|
| Base Clock | 1.44 GHz |
| Boost Clock | 1.77 GHz |
| Memory | |
| Memory | 16 GB |
| Memory Type | GDDR6 |
| Bandwidth | 200 GB/s |
| Interface Width | 128-bit |
| Interconnect & I/O | |
| PCIe | 4.0 x8 |
| Power & Thermal | |
| TDP | 60 W |
| Max power | 60 W |
| Cooling | Passive |
| Enterprise Features | |
| ECC Memory | Yes |
| SR-IOV | Yes |
| Sparsity | Yes |
| Physical & Media | |
| Width | 1-slot |
| General | |
| Form Factor | PCIe |
| Architecture | Ampere |
| Process Node | No verified data available |
| Launch Year | 2021 |
Datasheet & Resources
Data Provenance
Every figure traced to a source
Primary Source
- Document
- NVIDIA A2 Tensor Core GPU datasheet
- Publisher
- NVIDIA
- Published
- 2022-04-05
Data Quality
- Spec confidence
- Official
- Clock basis
- Boost
- Core precisions with figures
- 5 of 9
- Normalization
- All values in TFLOPS
Compare cloud providers offering on-demand GPU instances for AI training, inference, and HPC workloads.
Browse GPU Cloud ProvidersSimilar GPUs
Frequently Asked Questions
How many TFLOPS does the NVIDIA A2 have?
The NVIDIA A2 delivers 4.5 TFLOPS FP32 and 18 TFLOPS FP16 at peak. Flopper does not currently have a verified FP8 throughput figure for it.
What is the power consumption of the NVIDIA A2?
The NVIDIA A2 has a TDP (Thermal Design Power) rating of 60 watts.
How much memory does the NVIDIA A2 have?
The NVIDIA A2 is equipped with 16 GB of memory with 200 GB/s of memory bandwidth.
What architecture is the NVIDIA A2 based on?
The NVIDIA A2 is based on the Ampere architecture, launched in 2021.
Stay Updated on GPU Releases
Get notified when new GPUs are added or specifications are updated.
No spam, unsubscribe anytime.