Compute Comparison
vs
All comparisons →

Ampere

NVIDIA A40

wins

1

The NVIDIA A40 offers 48GB GDDR6 at a lower price point than the A100.

Large model inferenceProfessional renderingMulti-tenant AI

Ampere

NVIDIA A100 80GB

wins

4

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

Performance comparison

Visual bars — winner highlighted

NVIDIA A40

Metric

NVIDIA A100 80GB

FP16 TFLOPS

150 T
312 T

Bandwidth

696 GB/s
2.0k GB/s

VRAM

486 GB
802 GB

TDP (lower=better)

300 W
400 W

TFLOPS/watt

0.50
0.78

GB/s per watt

2.32
5.10

Full specification table

SpecNVIDIA A40NVIDIA A100 80GB
ArchitectureAmpereAmpere
VRAM48 GB GDDR680 GB HBM2e
VRAM typeGDDR6HBM2e
Memory bandwidth696 GB/s2,039 GB/s
FP16 TFLOPS150 TFLOPS312 TFLOPS
TDP300 W400 W
TFLOPS/watt0.50 T/W0.78 T/W
GB/s per watt2.32 GB/s/W5.10 GB/s/W
NVLinkNoYes
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 A40

FP16 TFLOPS150 TFLOPS
TDP300 W
TFLOPS/watt0.500 T/W
GB/s per watt2.32 GB/s/W

NVIDIA A100 80GB

FP16 TFLOPS312 TFLOPS
TDP400 W
TFLOPS/watt0.780 T/W
GB/s per watt5.10 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 A40 for:

  • Large model inference
  • Professional rendering
  • Multi-tenant AI

Choose NVIDIA A100 80GB for:

  • ML training
  • Inference at scale
  • HPC