AI, Machine Learning & Deep Learning
AI, machine learning and deep learning on Bare Metal servers
Given their potential, recent artificial intelligence (AI) innovations are re-shuffling the deck in the cloud market. To help you keep up with AI trends, our dedicated servers are equipped with high-performance NVIDIA GPU cards that come with advanced features. These give you the resources to deploy your data inference, deep learning and machine learning projects.
A new generation of servers, equipped to meet the new challenges of AI
Our Scale-GPU and HGR-AI servers are equipped with NVIDIA L4 and L40S GPU cards, respectively. They have large storage capacities — up to several terabytes, with a private network of 25 to 50 Gbps included. With these resources, you can process vast amounts of data to develop innovative AI solutions.
Raw performance
Our servers are equipped with the best hardware, and offer raw performance for high workloads. Your server’s resources are fully allocated to you, and the infrastructure is not shared with any other users. Furthermore, to process deep learning applications, our servers are equipped with graphic cards that accelerate workflows on your AI projects.
Security for your data
Data valorization and transformation are at the heart of all artificial intelligence projects. You then need to be able to control and secure your data across the entire usage chain. By opting for OVHcloud, you can ensure that your data is hosted on an infrastructure that respects data privacy, and is fully compliant with the GDPR. As a European cloud provider, we are not subject to laws like the US CLOUD Act.
Open-source solutions
Our solutions are based on open-source technology, so that you can simplify the way you use your services, and maintain control over your projects. We ensure compatibility with market standards, and full reversibility for your data. The use of open-source technology on both the server and application sides also represents significant cost-saving, with zero compromises.
Additional solutions
We offer a range of solutions tailored for different uses, as well as a portfolio of additional and complementary services. This way, your project evolves on a solid infrastructure, and you can add everything you need to it: Public Cloud, Hosted Private Cloud, dedicated servers, and much more.
Use case: a deep learning project
Description
Machine learning frameworks such as TensorFlow or PyTorch rely on very quick data throughput and GPU processors to deliver optimal results. If higher volumes of data can be analysed in a defined timeframe, machine learning tools are capable of making better decisions. This is why it is important to have high computing power, very fast disk access (IO), and GPUs designed for AI. Our Bare Metal servers ensure the best performance possible to meet these challenges.
Recommended servers for this use case:
- Scale-GPU servers: equipped with NVIDIA L4 cards; 30.3 TFlops FP32, 120 TFlops TF32, 242 TFlops FP16, 485 TFlops FP8.
- HGR-AI servers: equipped with NVIDIA L40S cards; 91.6 TFlops FP32, 366 TFlops TF32, 733 TFlops FP16, 1466 TFlops FP8.
Options and services included
- 25 Gbps (Scale) or 50 Gbps (High Grade) private bandwidth included.
- Up to 6 x 7.68 TB NVMe (Scale) or 4 x 15.36 TB NVMe hot swap (High Grade) storage — (optional).
- OVHcloud Link Aggregation (OLA) is available, and enabled by default (at no extra cost).
