NVIDIA A100 40GB PCIe vs NVIDIA A40, AI & Machine Learning Comparison
NVIDIA A100 40GB PCIe wins 8 of 9 benchmarks, averaging 79.8% faster.
Both cards' numbers are anchored estimates calibrated against our measured cards, pending first-party measurement. Treat small gaps as ties.
What the numbers say
The gap is widest in Z-Image Turbo, where NVIDIA A100 40GB PCIe leads by 188% (9.3 vs 3.225 images/min); the closest fight is Qwen3-4B (23% apart); VRAM decides part of this one: NVIDIA A40 runs 18 of our 12 AI workloads while the other card runs 8, models that don't fit score zero.
Benchmark results head-to-head
| Benchmark | NVIDIA A100 40GB PCIe | NVIDIA A40 | Difference |
|---|---|---|---|
| Qwen3-4B tok/s | 197.27 | 160.79 | +23% |
| Llama-3.1-8B tok/s | 154.22 | 106.18 | +45% |
| Qwen2.5-Coder-14B tok/s | 84.11 | 57.85 | +45% |
| Qwen3-32B tok/s | 41.5 | 26.89 | +54% |
| Llama 3.3 70B tok/s | 0 | 0 | n/a |
| Stable Diffusion XL images/min | 16.8 | 9.2 | +83% |
| FLUX.1 dev images/min | 3.964 | 1.95 | +103% |
| Z-Image Turbo images/min | 9.3 | 3.225 | +188% |
| Wan 2.2 5B (720p) frames/s | 0.61 | 0.31 | +97% |
Whole-job comparison
How long each card takes to finish a complete pipeline, not just one model. NVIDIA A100 40GB PCIe is faster on 3 of 3; NVIDIA A40 on 0.
| Workflow | NVIDIA A100 40GB PCIe | NVIDIA A40 | Difference | Cost per run |
|---|---|---|---|---|
| 24-frame storyboard | 3.2 min | 10.2 min | NVIDIA A100 40GB PCIe 3.18x faster | $0.027 vs $0.060 |
| Full codebase review | 11.9 min | 17.3 min | NVIDIA A100 40GB PCIe 1.45x faster | $0.101 vs $0.101 |
| Short social clips | 14.7 min | 33.4 min | NVIDIA A100 40GB PCIe 2.27x faster | $0.125 vs $0.195 |
Renting by the hour, NVIDIA A100 40GB PCIe finishes 2 of 3 cheaper. The quicker card is not automatically the cheaper way to get the work done.
Cost to rent
| Card | Per hour |
|---|---|
| NVIDIA A100 40GB PCIe | $0.508 |
| NVIDIA A40 | $0.350 |
NVIDIA A40 is 1.45x cheaper per hour. Cheapest on-demand rate we see across RunPod and Vast.
Specifications compared
| NVIDIA A100 40GB PCIe | NVIDIA A40 | |
|---|---|---|
| VRAM | 40GB | 48GB |
| Transistors | 54,200M | 28,300M |
| Die size | 826 mm² | 628.4 mm² |
| Process node | 7 nm | 8 nm |
| Transistor density | 65.6 M/mm² | 45 M/mm² |
| Architecture | Ampere | Ampere |
| Memory bandwidth | 1555 GB/s | 696 GB/s |
| Boost clock | 1,410 MHz | 1,740 MHz |
| TDP | 250 W | 300 W |
| Launch MSRP | $10,000 | $5,000 |
| Release | 2020-06-22 | 2020-10-05 |
FAQ
- Which is better for ai & machine learning: NVIDIA A100 40GB PCIe or NVIDIA A40?
- NVIDIA A100 40GB PCIe performs better for ai & machine learning, winning 8 of 9 benchmarks in our suite with an average 79.8% advantage.
- What are the main hardware differences between NVIDIA A100 40GB PCIe and NVIDIA A40?
- NVIDIA A100 40GB PCIe has 40GB VRAM and a 250W TDP, while NVIDIA A40 has 48GB VRAM and a 300W TDP.
- Does VRAM matter more than speed between NVIDIA A100 40GB PCIe and NVIDIA A40?
- For AI, yes, NVIDIA A40 fits 2 more of our 12 workloads than NVIDIA A100 40GB PCIe. A model that exceeds VRAM doesn't run slower, it doesn't run at all, so the card that fits the model wins that workload outright.
- Where is the biggest performance difference between NVIDIA A100 40GB PCIe and NVIDIA A40?
- Z-Image Turbo: NVIDIA A100 40GB PCIe leads by roughly 188% (9.3 vs 3.225 images/min) in our testing.
NVIDIA A100 40GB PCIe full review · NVIDIA A40 full review · All AI & Machine Learning rankings