AWS In stock
USA 3 configs 1x-16x
On-Demand
from $0.90
Nvidia K80
Legacy Kepler hardware, not viable for modern workloads.
Last update 11 minutes ago
Aggregating historical prices...
Key Specifications
K80 Pricing and Availability
The K80 is available at 2 cloud providers for $0.90/hr per GPU.
Database Mart
USA 4 configs 2x
Reserved
from $0.08
Heads up: We do our best to keep these specs & prices accurate. However, cloud costs may fluctuate based on region, usage, and other factors not listed here. These are estimates based on common setups and are for informational purposes only. Always verify current rates & exact specs with the provider before provisioning.
Frequently Asked Questions
Why choose the K80?
Dual-GPU card with 24GB total VRAM (12GB per die). Extremely low cost. Only suitable for basic GPU experimentation and legacy CUDA applications.
When is the K80 not a good fit?
Kepler architecture, 5+ generations behind. No FP16 Tensor Cores. Only viable for basic CUDA learning exercises or legacy applications.
Are K80 prices going up or down?
The median on-demand price has held steady around $0.90/hr per GPU across providers.
What size AI models can the K80 run?
With 12GB of VRAM, the K80 is best for 7B-class models in 4-bit or 8-bit quantized form, and smaller models in FP16.
How much VRAM does the K80 have?
The K80 has 12GB of VRAM. Multi-GPU setups increase total memory, but that memory is not automatically pooled across GPUs.
What is the K80's memory bandwidth?
The K80 has 240 GB/s of memory bandwidth. Higher bandwidth helps with faster data transfer between GPU memory and compute cores.
What data types does the K80 support?
The K80 supports 2 precision formats. Training: FP32. Scientific: FP64.
Does the K80 support NVLink?
No. The K80 is a PCIe-only GPU with no NVLink, so it is better suited to single-GPU inference and smaller-scale workloads than large distributed training jobs.
How much does the K80 cost per hour?
K80 pricing currently ranges from $0.08/hr to $0.90/hr per GPU, depending on the provider, instance type, and billing model.
How much does the K80 cost per month?
At 720 hours per month, one K80 can cost between $54.50 to $648.00 per month, depending on the provider. Reserved and spot pricing can lower that further.
Which cloud providers offer the K80?
The K80 is available from 2 cloud providers: Amazon Web Services, Database Mart.
Can I rent the K80 in the cloud?
Yes. We currently track 7 K80 listings across 2 cloud providers:
Billing type Listings Avg $/GPU/hr On-demand 3 $0.90/hr Reserved 4 $0.08/hr
Technical Specifications
| Tesla K40 | Tesla K80 | |
|---|---|---|
| Peak double-precision floating point performance (board) | 1.43 Tflops | 1.87 Tflops |
| Peak single-precision floating point performance (board) | 4.29 Tflops | 5.6 Tflops |
| GPU | 1 x GK110B | 2 x GK210 |
| CUDA Cores | 2,880 | 4,992 |
| Memory size per board (GDDR5) | 12 GB | 24 GB |
| Memory bandwidth for board (ECC off) | 288 Gbytes/sec | 480 Gbytes/sec |
| Architecture features | SMX, Dynamic Parallelism, Hyper-Q | SMX, Dynamic Parallelism, Hyper-Q |
| System | Servers and workstations | Servers |
Source: official Nvidia K80 datasheet.
Alternatives to Nvidia K80
Last updated