Compute Comparison
vs
All comparisons →

Ampere

NVIDIA A100 80GB

wins

2

The A100 remains one of the most widely available data center GPUs. Its broad provider support and mature software ecosystem make it a reliable choice for ML training and inference.

ML trainingInference at scaleHPC

Ampere

NVIDIA A100 40GB

wins

0

The A100 40GB is the more affordable sibling of the A100 80GB.

ML trainingInferenceCost-effective HPC

Performance comparison

Visual bars — winner highlighted

NVIDIA A100 80GB

Metric

NVIDIA A100 40GB

FP16 TFLOPS

312 T
312 T

Bandwidth

2.0k GB/s
1.6k GB/s

VRAM

802 GB
402 GB

TDP (lower=better)

400 W
400 W

TFLOPS/watt

0.78
0.78

GB/s per watt

5.10
3.89

Full specification table

SpecNVIDIA A100 80GBNVIDIA A100 40GB
ArchitectureAmpereAmpere
VRAM80 GB HBM2e40 GB HBM2e
VRAM typeHBM2eHBM2e
Memory bandwidth2,039 GB/s1,555 GB/s
FP16 TFLOPS312 TFLOPS312 TFLOPS
TDP400 W400 W
TFLOPS/watt0.78 T/W0.78 T/W
GB/s per watt5.10 GB/s/W3.89 GB/s/W
NVLinkYesYes
Release year20202020

Live cloud pricing

On-demand $/hr across providers — updated in real time

Loading prices…

Power efficiency analysis

TFLOPS/watt and GB/s/watt — critical for data center TCO

NVIDIA A100 80GB

FP16 TFLOPS312 TFLOPS
TDP400 W
TFLOPS/watt0.780 T/W
GB/s per watt5.10 GB/s/W

NVIDIA A100 40GB

FP16 TFLOPS312 TFLOPS
TDP400 W
TFLOPS/watt0.780 T/W
GB/s per watt3.89 GB/s/W

TFLOPS/watt measures compute efficiency — how much AI throughput you get per watt of power consumed. For data centers with PUE of 1.2–1.5, a 10% improvement in TFLOPS/watt translates directly to lower electricity costs and cooling requirements. GB/s/watt measures memory bandwidth efficiency, which is the binding constraint for memory-bound LLM inference workloads.

When to choose each GPU

Choose NVIDIA A100 80GB for:

  • ML training
  • Inference at scale
  • HPC

Choose NVIDIA A100 40GB for:

  • ML training
  • Inference
  • Cost-effective HPC