Product details
Key Features
- Rent H100 80GB GPU Cloud Instance
- Suitable for AI/ML, Deep Learning, Large Language Models (LLM) applications
- Instant access to NVIDIA H100 80GB GPUs (once reserved and start date is finalized)
- Four Nodes – HGX H100 SXM5 NVLInk Platform. Each node features:
- 8 x NVIDIA H100 80GB GPU
- Up to 2TB ECC Reg Memory
- Flexible storage options
- Pre-installed OS and AI packages
- Custom software setup for your application (available on request)
- Pricing is for per 32-GPU (4 nodes) per month
- For multiple nodes, adjust qty and add to cart
What are Dihuni’s GPU Cloud Offerings?
Dihuni offers monthly, 6-monthly, annual and 3 year cloud infrastructure access for world’s leading GPUs. You can order GPU instances that you need right now and renew as required on a subscription basis. Subscriptions can be manually renewed as required.
Why would I want to access GPUs in the Cloud?
There are multiple reasons why GPU cloud access can be suitable for you:
- Budget – if you don’t have the budget for fully assembled on-premise GPU systems, you can spend a fraction of the price of systems and run your programs for as long as your budget allows
- Scale – You can scale from 8 GPU per node to thousands of GPUs in a cluster without worrying about purchasing expensive hardware
- Zero maintenance – with DIhuni’s cloud, you don’t have to worry about setting up hardware or installing operating system or AI packages. All that work is done by us so you can just focus on developing your AI/ML/LLM applications.
- Use while waiting for your on-premise hardware – GPU systems can have a long lead time. If you have ordered long lead-time GPU systems from Dihuni, you can get a head-start using GPUs in the cloud to fill in the gap until your systems arrive at your door.
How does it Work?
Dihuni has collaborated with Qubrid, world’s leading GPU Cloud platform company to bring you an easy to use interface to deploy your cloud instance and program in a Jupyter notebook or another programming environement with your choice of AI packages including Tensorflow, Keras etc. All you have to do is login to the Qubrid Cloud platform using a web browser and access your GPU instance with a click of a button. To try this out on a free CPU instance, please login to Qubrid platform.
How long would it take for my H100 instance to be activated?
Current eta is 3-4 weeks. Sometimes you can get your instance in one week but if the nodes are being used by other customers, it can take a few weeks to get you the access. Please submit your deployment requirements and time-frame with your order or quote request and we will get back to you with the estimated time of activation.