NVIDIA · Ampere · 2020
NVIDIA A100 40GB. 40 GB HBM2, 312 TFLOPS FP16 tensor.
Ampere with 40 GB of HBM2: A100 compute for models that fit in 40 GB, for less per hour.
A100 40GB on demand
from$1.99/hr per GPU
- Availability
- Available now · 5 regions
- By region and size
- $1.99 to $4.15/hr
- Billing
- By the second on most GPUs, a few by the minute
- 8+ GPUs or a month
- Launch any size yourself, and prepay 1 to 12 months when you launch an instance. Custom quotes on request.
The machine
40 GB of HBM2. At 1.6 TB/s.
- Memory
- 40 GB
- HBM2
- Bandwidth
- 1.6 TB/s
- Memory to compute
- FP16 tensor
- 312
- TFLOPS
- Per node
- 8 GPUs
- up to 400 W each
Workloads
What teams run on the A100 40GB. Each one opens the launch form, sized.
Full specs
NVIDIA A100 40GB specifications.
Memory
- Memory
- 40 GB
- Type
- HBM2
- Bandwidth
- 1.6 TB/s
Compute
- FP32
- 19.5 TFLOPS
- FP16 tensor
- 312 TFLOPS
- INT8
- 624 TOPS
- Tensor cores
- 432
- CUDA cores
- 6,912
Power and form
- TDP
- up to 400 W
- Architecture
- Ampere
- Released
- 2020
Node
- GPUs per node
- 8
- Maker
- NVIDIA
Alternatives
Alternatives to the A100 40GB. Same account, different price and performance.
| GPU | Memory · FP16 tensor | Stands out | Price gap | From |
|---|---|---|---|---|
| NVIDIA A100 40GBThis GPU | 40 GB HBM2 · 312 TFLOPS | $1.99/hr | ||
| NVIDIA A100 80GB | 80 GB HBM2e · 312 TFLOPS | +40 GB memory | -13% | $1.73/hr |
| NVIDIA L40S | 48 GB GDDR6 · 362 TFLOPS | -43% | $1.14/hr | |
| AMD MI250X | 128 GB HBM2e · 383 TFLOPS | +88 GB memory | Not on demand | |
| NVIDIA RTX 5090 | 32 GB GDDR7 · 209.5 TFLOPS | -33% FP16 tensor | Not on demand |
Questions
Renting the A100 40GB. What buyers ask first.
Other GPUs
Browse every GPU by architecture.
NVIDIA Blackwell
NVIDIA Hopper
NVIDIA Ampere
NVIDIA Ada Lovelace
NVIDIA Volta