Rent NVIDIA Tesla P40 GPU by the hour
The NVIDIA Tesla P40 is a Pascal-architecture datacenter GPU equipped with 24 GB of VRAM and compute capability 6.1. It provides about 9.4 TFLOPS of single-precision compute and about 257 GB/s of memory bandwidth within a 250 W thermal envelope. On ONIE, you can rent this GPU with per-hour billing, SSH access, and pre-configured templates for PyTorch/CUDA, vLLM, Ollama, and ComfyUI.
Specifications
VRAM per GPU
Architecture
Compute capability
FP32 compute
Memory bandwidth
Power limit
Values reported by the infrastructure provider for machines currently listed; real performance varies by machine.
Available configurations
Live stock and prices, updated continuously. The final price is confirmed before you order.
Loading live prices…
What the Tesla P40 is good for
Quantized LLM Inference
Run local language models that require up to 24 GB of memory using ready templates like Ollama or vLLM.
Image Generation Workflows
Host ComfyUI pipelines that benefit from generous VRAM capacity for loading multiple checkpoints and control nets.
Legacy CUDA Development
Build, test, and maintain PyTorch/CUDA applications targeting Pascal architecture and compute capability 6.1.
Memory-Bound Batch Processing
Execute datacenter workloads that prioritize fitting large datasets into 24 GB of VRAM over raw compute speed.
Is it the right GPU for you?
Choose the Tesla P40 if your primary requirement is 24 GB of VRAM for memory-hungry inference or development tasks where older Pascal compute is sufficient. Opt for a newer or larger GPU if you need higher memory bandwidth or modern tensor acceleration for heavy training, or a smaller GPU if your workload uses little memory.
Compare with similar GPUs
Frequently asked questions
Is the Tesla P40 suitable for running large language models?
The 24 GB VRAM capacity allows the Tesla P40 to hold moderately sized, quantized language models in memory using templates like Ollama. However, because its memory bandwidth is about 257 GB/s and it uses the older Pascal architecture, token generation will be slower than on newer datacenter GPUs.
What software compatibility should I expect with compute capability 6.1?
Compute capability 6.1 supports standard single-precision workloads at about 9.4 TFLOPS, which works well with general PyTorch/CUDA templates. Some modern machine learning libraries or features that require newer hardware instructions may not run on the Pascal architecture.
How do I connect to and set up my Tesla P40 instance on ONIE?
You can deploy the instance using ready templates for PyTorch/CUDA, vLLM, Ollama, or ComfyUI. Direct SSH access is included so you can manage files and custom environments from your terminal.
How does billing work for renting the Tesla P40?
Instances on ONIE use per-hour billing, allowing you to pay only for the duration your datacenter GPU instance is active.
Rent the Tesla P40 by the hour
Create an account, choose a configuration and connect over SSH. Billing stops when you delete the GPU.
Rent Tesla P40