Flopper.io Flopper.io

NVIDIA A100 SXM4 40GB vs NVIDIA A40 48GB

The A100 delivers 2.1x the FP16 throughput of the A40 (312 vs 150 TFLOPS dense). The A40 counters with 8 GB more memory (48 GB vs 40 GB).

A100: Ampere, 2020 A40: Ampere, 2021
FP16 dense lead
2.1x
A100: 312 vs 150 TFLOPS
Memory
40 vs 48 GB
8 GB more for the A40
Bandwidth
1.6 TB/s vs 696 GB/s
A100 moves data faster
TDP
400 W vs 300 W
A40 draws 100 W less
See rental pricing from $0.35/hr Download comparison sheet

Share this page

NVIDIA A100 SXM4 40GB vs NVIDIA A40 48GB

A post with this link shows the card on the Image tab as its preview.

Post on X Share on LinkedIn
 Printable version →

NVIDIA A100 SXM4 40GB

Leads on FP16 Full specs →

What is the NVIDIA A100?

The A100 is a GPU from NVIDIA on the Ampere architecture, an accelerator for AI training, inference and high-performance computing. It is an SXM module, the mezzanine format used in dense multi-GPU servers. A100 full specifications →

Can I rent the A100 in the cloud?

Yes. Cloud providers tracked by Flopper rent the A100 by the hour, and the pricing page compares their current rates and the price history side by side. A100 rental prices →

How does the A100 compare with other accelerators?

This page sets the A100 against the A40. The A100 is the older generation of the two. Both are NVIDIA parts; the A100 is an SXM module while the A40 is a PCIe card. The compare tool builds the same comparison for any two parts in the catalogue. All GPU comparisons →

Where do the A100 figures come from?

Every A100 figure on this page traces to NVIDIA's own datasheet or specification table. How Flopper verifies specifications →

NVIDIA A40 48GB

Full specs →

What is the NVIDIA A40?

The A40 is a GPU from NVIDIA on the Ampere architecture, an accelerator for AI training, inference and high-performance computing. It is a PCIe add-in card, so it fits standard servers and workstations. A40 full specifications →

Can I rent the A40 in the cloud?

Yes. Cloud providers tracked by Flopper rent the A40 by the hour, and the pricing page compares their current rates and the price history side by side. A40 rental prices →

How does the A40 compare with other accelerators?

This page sets the A40 against the A100. The A40 is the newer generation of the two. Both are NVIDIA parts; the A40 is a PCIe card while the A100 is an SXM module. Flopper also compares it head to head with the SC11 FP300, across every published precision, memory and power. A40 vs SC11 FP300 →

Where do the A40 figures come from?

Every A40 figure on this page traces to NVIDIA's own datasheet or specification table. How Flopper verifies specifications →

Specifications

A100
A40
Architecture
AmpereAmpere
Launch Year
20202021
Form Factor
SXMPCIe
Memory
40 GB48 GB
Memory Bandwidth
1.6 TB/s696 GB/s
TDP
400 W300 W
Process Node
7nm—

Performance (TFLOPS)

A100
A40
9.7 TFLOPS No verified data available
20 TFLOPS 37 TFLOPS
156 TFLOPS
312 TFLOPS sparse
75 TFLOPS
150 TFLOPS sparse
312 TFLOPS
624 TFLOPS sparse
150 TFLOPS
299 TFLOPS sparse
312 TFLOPS
624 TFLOPS sparse
150 TFLOPS
299 TFLOPS sparse
No verified data available No verified data available
FP6
No verified data available No verified data available
No verified data available No verified data available
624 TOPS
1,248 TOPS sparse
299 TOPS
599 TOPS sparse

Figures are dense. Where a vendor publishes a figure it sits under the dense one and never decides a winner. Hover any precision for its definition, or read the GPU glossary.

FLOPS by Precision

What actually differs

The A40 is the newer part: Ampere (2021) against the A100's Ampere (2020).

The A40 carries 8 GB more memory (48 GB against 40 GB), so larger models fit on one GPU before you shard. Bandwidth is 1.6 TB/s against 696 GB/s, which sets the ceiling on decode speed.

Power is 400 W against 300 W, and at FP32 the A40 does more work per watt. Weight the FP8 and FP16 rows for LLM work and the FP64 row for HPC.

New to the terms? The glossary explains FP64, FP32, TF32, BF16, FP16, INT8, Sparsity and TFLOPS. For how the generations relate, see NVIDIA architectures explained.

Frequently asked questions

Is the A100 faster than the A40?

At FP16 precision the A100 reaches 312 TFLOPS dense against 150 TFLOPS for the A40. The performance table on this page lists every published precision for both GPUs.

Which should you choose, the A100 or the A40?

Choose the A100 for training and high-throughput inference: it has 2.1x the FP16 throughput of the A40. Choose the A40 when the model has to fit on one GPU: it carries 8 GB more memory. The A100 draws 100 W more power, so budget for cooling and power delivery before you commit. Packaging differs too: the A100 is an SXM module while the A40 is a PCIe card, so check what your servers accept.

Which has more memory, the A100 or the A40?

The A40 carries 48 GB of memory versus 40 GB for the A100. Memory bandwidth is 1.6 TB/s for the A100 and 696 GB/s for the A40.

How much power do the A100 and the A40 draw?

The A100 is rated at 400 W TDP and the A40 at 300 W. On FP32 throughput per watt, the A40 is the more efficient part.

How much do the A100 and the A40 cost to rent?

Live cloud listings tracked by Flopper price the A100 from $0.45 per GPU-hour across 9 providers, and the A40 from $0.35 per GPU-hour across 3 providers. The rental section on this page links to every provider, rate and the price history for each GPU.

Where to Rent

NVIDIA A100 SXM4 40GB

Rent from

$0.45 / GPU-hr

Lowest of 14 live listings across 9 providers · median $2.05

See all A100 rental prices →

Price history and every provider on the pricing page. Some links are affiliate links.

NVIDIA A40 48GB

Rent from

$0.35 / GPU-hr

Lowest of 4 live listings across 3 providers

See all A40 rental prices →

Price history and every provider on the pricing page. Some links are affiliate links.

Get Comparison Updates

New GPUs added weekly. Be the first to see how they compare.

Open in Full Compare Tool → A100 full specs A40 full specs Estimate cluster cost in the calculator →

© 2026 Flopper.io - Compare the GPUs Powering AI