Leaderboard Β· All comparisons Β· LynxBenchAI 26Q3 Index

NVIDIA A100 80GB PCIe vs NVIDIA GeForce RTX 4080 SUPER for AI

Device pages: NVIDIA A100 80GB PCIe Β· NVIDIA GeForce RTX 4080 SUPER

On the LynxBenchAI 26Q3 Index, the NVIDIA A100 80GB PCIe scores 1.01x the NVIDIA GeForce RTX 4080 SUPER overall (Training 1.33x, Inference 1.06x, Compute 0.54x). It leads on Training and Inference and trails on Compute.

Scores on the LynxBenchAI 26Q3 Index

Ratio = the A100 80GB PCIe's score divided by the RTX 4080 SUPER's
ScoreRatioA100 80GB PCIeRTX 4080 SUPER
Overall (GT)1.01x403.38398.74
Training1.33x387.30291.00
Inference1.06x584.07551.54
Compute0.54x201.12370.70
  • Runs on the RTX 4080 SUPER only: FP8 inference.

When to Choose Which

  • Choose the A100 80GB PCIe for training (1.33x the RTX 4080 SUPER).
  • Level on inference: the two are within 10% (1.06x).
  • Choose the RTX 4080 SUPER for raw compute (1.84x the A100 80GB PCIe).

From the LynxBenchAI 26Q3 Index category scores, which are not raw throughput; a lead under 10% counts as level.

Per Model and Precision

LynxBenchAI score per model and precision; ratio = A100 80GB PCIe / RTX 4080 SUPER
ModelPrecisionRatioA100 80GB PCIeRTX 4080 SUPER
compute
Dense MatMulBF161.19x418.74352.69
Dense MatMulFP161.11x384.97347.27
Dense MatMulFP321.00x317.87318.83
Dense MatMulFP6424.21x117.844.87
Sparse MatMulFP320.15x92.16599.74
Sparse MatMulFP640.33x163.89495.92
inference
DistilBERTFP161.20x1102.80921.58
DistilBERTFP8n/anot run645.88
DistilBERTINT81.05x531.81506.40
DistilGPT2FP161.70x1390.64817.72
DistilGPT2FP8n/anot run504.17
DistilGPT2INT81.37x630.82458.99
Flan-T5 SmallFP161.16x1411.061216.48
Flan-T5 SmallFP8n/anot run763.76
Flan-T5 SmallINT81.42x672.11472.69
GraphGPS (Peptides)FP162.05x525.67256.57
GraphGPS (Peptides)FP8n/anot run128.56
GraphGPS (Peptides)INT82.04x260.48127.77
MobileNetV3FP161.22x1472.071205.33
MobileNetV3INT81.39x459.72330.60
training
DistilBERTBF161.64x639.53389.61
DistilBERTFP321.29x736.08571.22
DistilGPT2BF161.67x343.98205.47
DistilGPT2FP321.56x609.72390.78
Flan-T5 SmallBF161.54x336.21218.55
Flan-T5 SmallFP321.37x579.06423.07
GraphGPS (Peptides)BF161.27x214.16168.65
GraphGPS (Peptides)FP321.34x212.64159.03
MobileNetV3BF160.80x173.31215.50
MobileNetV3FP321.07x346.28323.35

Related Comparisons

Each figure is the first device's overall score divided by the second's.

Measure your own machine: pip install lynxbench-ai Β· Install guide β†’

How to Read These Results

A LynxBenchAI score is sustained throughput on a fixed catalogue of PyTorch models, measured after a warm-up at a batch size raised until throughput stops improving, so it is what the machine holds under load, not its best instant. Training, Inference and Compute are three separate scores; GT combines them (Training and Inference weigh 0.4 each, Compute 0.2).

Every run uses the stock, publicly released PyTorch stack, so a result is broadly indicative of the out-of-the-box PyTorch experience on that hardware, and so of custom PyTorch models too. It does not run your own model, and it is not a vendor-tuned peak like an MLPerf submission.

Scores compare only within one Index: the 26Q3 Index fixes the models, precisions, correctness thresholds and scoring, and a later Index may change them. Each test is one timed window, the scores have no 0-100 scale, and a precision a device cannot run scores zero rather than being skipped.

LynxBenchAI is built by TechnoLynx, which the European Commission engaged as an independent expert advisor on AI performance and benchmarking for AI Gigafactories under the EuroHPC Joint Undertaking (case study).

Questions

Is the NVIDIA A100 80GB PCIe faster than the NVIDIA GeForce RTX 4080 SUPER for AI?
On the LynxBenchAI 26Q3 Index, the NVIDIA A100 80GB PCIe scores 1.01x the NVIDIA GeForce RTX 4080 SUPER overall (Training 1.33x, Inference 1.06x, Compute 0.54x). It leads on Training and Inference and trails on Compute.
Should I choose the NVIDIA A100 80GB PCIe or the NVIDIA GeForce RTX 4080 SUPER?
Choose the A100 80GB PCIe for training (1.33x the RTX 4080 SUPER). Level on inference: the two are within 10% (1.06x). Choose the RTX 4080 SUPER for raw compute (1.84x the A100 80GB PCIe). These follow the LynxBenchAI 26Q3 Index category scores, which are not raw throughput.
Where is the difference between the NVIDIA A100 80GB PCIe and the NVIDIA GeForce RTX 4080 SUPER largest?
On the LynxBenchAI 26Q3 Index, the ratio is highest on Dense MatMul at FP64 (Compute), where the NVIDIA A100 80GB PCIe scores 24.21x the NVIDIA GeForce RTX 4080 SUPER, and lowest on Sparse MatMul at FP32 (Compute), at 0.15x.
Which precisions run on only one of the NVIDIA A100 80GB PCIe and the NVIDIA GeForce RTX 4080 SUPER?
On the LynxBenchAI 26Q3 Index, only the NVIDIA GeForce RTX 4080 SUPER runs FP8 inference; a precision a device cannot execute scores zero.
What does this result tell me about my own PyTorch model?
Every run uses the stock, publicly released PyTorch stack, so a result is broadly indicative of the out-of-the-box PyTorch experience on that hardware, and so of custom PyTorch models too. It does not run your own model, and it is not a vendor-tuned peak like an MLPerf submission.

The Personal Edition submits your benchmark results to TechnoLynx's servers as part of participating in the public leaderboard. If that isn't acceptable for your use case, contact TechnoLynx about Press, Pro, or Enterprise licenses.

Get Started