High Performance Computing (HPC)
High performance computing (HPC) on a bare-metal server
With high-performance computing solutions, you can address the challenges of high-performance computing and massive data processing. Robust, high-performance, purpose-built servers are required to meet the power, memory, and storage requirements. Our Bare Metal dedicated servers offer a concrete solution to these challenges.
Create your high-performance computing architecture with OVHcloud servers
A high-performance computing architecture is based on three fundamental aspects: computing power, network capacity, and storage capacity. Balancing the server components is essential to ensure stable computing workloads, and efficient processing. Our Bare Metal servers are equipped with the latest generation high-performance components, and have been designed to meet these challenges. In particular, the High Grade range offers raw performance, with a very high-bandwidth and high-resilience network. Scale your infrastructure over time by adding nodes to your cluster, or allocating additional resources (RAM, storage disks, bandwidth) to it.
Unrivalled computing power
Our servers are equipped with the latest-generation AMD EPYC, Intel Gold and dual processors, so they can deliver exceptional raw performance. For the most ambitious projects, our HGR-AI server integrates 4 NVIDIA V100S GPU cards to offer optimal computing performance and accelerate your HPC operations.
A balanced platform
With a memory capacity ranging from 128GB to 2TB per server, your applications no longer experience high latency due to disk access, and can work alongside one another without difficulty. Our servers are equipped with DDR4 ECC RAM, which guarantees performance and reliability.
A high-resilience cluster
Consolidate your resources reliably and securely with our high-speed private network, and guaranteed bandwidth of up to 50Gbit/s*. With the OVHcloud Link Aggregation feature, you are guaranteed high resilience.
*On the High Grade range.
High-performance storage
With storage included as SATA, SAS or NVMe drives, you can select the best level of performance and price to suit your needs. With our external storage solutions, you can expand, share and back up your data at your own pace, while remaining within your budget.
Description
Choose the best servers for your project from our Scale and High Grade server ranges. In just a few clicks, you can customise your infrastructure by choosing memory, storage, public and private bandwidth to suit your needs.
High Grade servers stand out because their disks can be hot-swapped — so you can increase your server’s storage capacity without experiencing any service interruptions.
In addition to the range of operating systems available for pre-installation, you can import a distribution or operating system of your choice with the Bring Your Own Image feature.
You also get the most powerful anti-DDoS protection to protect against distributed denial-of-service attacks — and you can add the OVHcloud Load Balancer to balance your workloads efficiently, regardless of their location.
Options and services included
- A range of memory and storage options.
- Up to 25Gbit/s guaranteed private bandwidth on the Scale range, and up to 50Gbit/s on the High Grade range (optional).
- OVHcloud Link Aggregation (OLA) available and enabled by default at no additional cost.
Description
Quickly add HGR-AI-1 servers, equipped with four NVIDIA V100S GPU cards, to your HPC cluster to boost your computing power.
The servers in these ranges include four 25Gbit/s network links each. You can dedicate them to communication between the nodes in your cluster, and get private bandwidth up to 100Gbit/s.
In addition to our Bare Metal servers, you can use our Cloud Disk Array and Enterprise File Storage solutions to combine performance and high availability. These are perfect for storing critical application data.
Options and services included
- A range of memory and storage options.
- Up to 50Gbit/s guaranteed private bandwidth on the High Grade range (optional).
- OVHcloud Link Aggregation (OLA) available and enabled by default at no additional cost.
Our recommended Bare Metal server ranges
A range of servers designed for complex, high-resilience infrastructures.
24-48 core processors and dual processors, max storage 24TB SSD NVMe, guaranteed private bandwidth of up to 25Gbps.
The most powerful servers, optimised for critical loads.
32-64 core dual processors, max storage of 92TB NVMe SSD or SAS SSD, up to 24 disks (hot-swappable), guaranteed private bandwidth of up to 50Gbps.
Solution optimised for AI and HPC operations.
Dual processors with 32 cores, up to 1.5TB SATA SSD storage and 8 disks (hot swap), guaranteed private bandwidth of up to 50Gbit/s.
What uses are there for high performance computing?
High performance computing is used across different business domains that share a common need — the need to perform many complex calculations simultaneously.
This technology is widely used in the field of science (e.g. physics or chemistry) and medicine. In industry as a whole, HPC solutions are used to deploy machine learning and artificial intelligence projects. Other domains, such as finance or entertainment platforms, harness the benefits of high-performance computing to process and assimilate the collected data streams.
Your questions answered
What is high performance computing?
High performance computing (HPC) is used for complex processing requirements, at speeds far higher than a standard computer. For the search and processing of high-volume data streams, a HPC architecture makes it possible to use high volumes of computing resources. This ensures that workloads are completed in a timely, balanced manner. The best components are selected for the performance they provide, and their ability to support each stage of processing. The RAM, network, storage and compute capacity must be balanced to ensure that the workflow does not slow down. A HPC solution can be deployed in a hybrid architecture for increased flexibility. You can do this by combining different products, like dedicated servers and cloud computing services — and this architecture can be deployed across multiple datacentres.
How does high performance computing work?
To deploy a HPC solution, you need to create a cluster of physical servers. This cluster groups together all of the hardware resources used for intensive computing. The balance between the performance and capacity of each resource is observed, particularly in relation to project requirements. From storage capacity and processing speed to transfer flow between servers, everything is calculated for optimal performance. Importantly, the agility offered by scalable resources and servers must also be able to meet the growing needs of applications, services and algorithms.
Server equipped with an Intel Xeon Gold 6226R dual processor (32c/64t @ 2.9GHz/3.9GHz)