NVIDIA A100 40GB SXM vs NVIDIA A100 80GB PCIe
A100 40GB SXM has 50% less memory than A100 80GB PCIe (40 GB vs 80 GB). Its memory bandwidth is 20% less (1.555 vs 1.935 TB/s). In the cloud, the median on-demand A100 40GB SXM price is $1.86 per GPU-hour against $4.78 for the A100 80GB PCIe. Per gigabyte of GPU memory, that is $0.047 vs $0.060 per hour.
Specifications
Manufacturer figures as recorded by Kovara. Higher memory and bandwidth are highlighted; neither is a benchmark of your workload.
| A100 40GB SXM | A100 80GB PCIe | |
|---|---|---|
| Manufacturer | NVIDIA | NVIDIA |
| Architecture | Ampere | Ampere |
| Memory | 40 GB HBM2 | 80 GB HBM2E |
| Memory bandwidth | 1.555 TB/s | 1.935 TB/s |
| Power (TDP) | 400 W | 300 W |
| Form factor | SXM | PCIe dual-slot |
Cloud prices today
Public offers Kovara tracks, normalised to US dollars per GPU-hour. Prices use on-demand terms where available; spot and reserved offers are on each GPU’s price pages. Lower is highlighted.
| A100 40GB SXM | A100 80GB PCIe | |
|---|---|---|
| Lowest price | $1.15 | $1.59 |
| Median price | $1.86 | $4.78 |
| Highest price | $1.99 | $7.35 |
| Median per GB of memory | $0.047 | $0.060 |
| Public offers (all terms) | 5 | 83 |
| Providers | 3 | 5 |
Where to rent each
Lowest tracked price per provider. Open a provider to see every offer, region and term.
Questions
Which is cheaper to rent, the A100 40GB SXM or the A100 80GB PCIe?
Right now the cheapest on-demand A100 40GB SXM Kovara tracks is $1.15 per GPU-hour and the cheapest A100 80GB PCIe is $1.59. Medians are $1.86 and $4.78. The cheaper GPU per hour is not always cheaper per job: a faster or larger GPU can finish sooner or need fewer units.
Which has more memory?
The A100 80GB PCIe has more memory: 80 GB against 40 GB. Memory decides which models and batch sizes fit on one GPU.
Are these benchmark results?
No. This page compares manufacturer specifications and public prices. Real performance depends on the model, precision, software and interconnect. Kova can help you estimate cost for a specific workload.