Rescale Bolsters Support for Containerization of HPC and AI/ML Workloads in the Cloud
See a hands-on demonstration of Rescale’s HPC container capabilities featuring the latest HPC and AI tools at the Nvidia GTC conference March 23, 2022
SAN FRANCISCO, March 21, 2022 — Rescale, the leader in high performance computing built for the cloud to accelerate engineering innovation, today announced broader and deeper support to run containers on any cloud and any specialized architecture, enabling companies to deploy their custom and open source science and engineering applications with greater simplicity and cost-performance on any hybrid cloud architectures.
Containers are popular for delivering workload portability and efficiency, but high performance computing use cases bring additional requirements on security and parallel processing. Rescale has long supported the use of containers to bundle applications and their dependencies needed to execute on the best possible architecture, and run those workloads with 800+ applications on Rescale’s always-current software catalog.
Containers simplify the development and deployment of custom software because they abstract the complexity of hybrid and multi-cloud deployments and dependencies that can arise from differences in infrastructure environments. Rescale integrates the leading container technologies to enable open source collaboration and proprietary applications development. Growth in home-grown application development, for example proprietary AI/ML models, has spurred further adoption of containers on Rescale to achieve increased engineering productivity and enhanced IT control.
Rescale will be demonstrating its enriched support for containers on March 23 at 12:00-12:25 PDT at Nvidia’s GTC conference, the leading showcase for new developments in AI. Register here to watch the demonstrations and ask questions of Rescale.
The benefits of using Rescale to run containers include:
- Multi-Cloud – Run your containerized applications on any cloud in any region;
- HPC-Optimized – Support more secure and parallel workloads while optimizing for best performance or cost efficiency;
- Integrated Experience – Integrate your container workloads alongside 800+ commercial & open source science and engineering applications.
“Organizations powering computational science and engineering with HPC can face daunting challenges of complexity,” said Adam McKenzie, CTO and Founder at Rescale. “Containers on Rescale help to extend the simplicity and full-stack cloud automation we bring our customers into containerized workloads. The rapid growth of AI/ML, open-source, and custom applications is driving a convergence with HPC. Rescale provides a bridge for these technologies to not just run applications, but supercharge them with the latest architectures whether you care about optimizing cost or speed while at the same time giving IT the security and control they require.”
Customers deploy their containerized workloads on Rescale in the cloud to access greater scale and better performance on alternate infrastructure. Containers also make it easier for IT operations to better support and manage their in-house applications seamlessly with their other commercial applications. From autonomous vehicle testing to industrial manufacturing optimization, Rescale customers are accelerating the training of AI/ML models and other proprietary applications using the Rescale platforms.
In-house HPC software development is often bespoke and run with large data sets across on-premises data centers or specific architectures. Containers make these same workloads portable, which is a big advantage when customers want improvements in performance and price. Containers deliver applications as microservices. These microservices maintain their lifecycle along with service-specific requirements of independent development, granular scaling, and patching and fault remediation.
Similarly, containers give HPC and AI/ML workloads the advantage of isolated management that includes scaling and development. Container scaling is a major advantage in HPC where workloads can spike in data processing requirements.
Rescale advances in container capabilities include:
- Multi-node high performance support for Singularity and Apptainer containers; Docker container support for single node applications;
- ARM architecture support for the latest Singularity versions (3.9.6);
- High-performance networking, such as InfiniBand using Open MPI;
- Container publishing via Rescale Templates for collaboration and standardization;
- NVIDIA GPU-optimized containers for seamless integration of world-class AI applications.
Rescale customer AGC, a global leader in glass, ceramic, chemicals and electronics materials manufacturing, founded in 1907, uses containers to support ML and Deep Learning (DL) capabilities using their cloud provider of choice. Containers give the company the portability and compatibility they need to run in-house applications anywhere. Read more about AGC here.
To learn more about Rescale and containers, please explore the following resources:
- Visit HPC Container Solutions on Rescale;
- Register for NVIDIA’s GTC event to attend a live demonstration of the latest AI technologies using containers on Rescale (3/23 at 12:00 PDT);
- Schedule a hands-on demonstration here;
- Get started with a Bring-your-own container tutorial on Rescale.
Backed by investors like Jeff Bezos, Sir Richard Branson, Microsoft, Samsung and NVIDIA, Rescale is accelerating engineering breakthroughs in a wide range of industries, making it possible for scientists and engineers to do more, faster.
Rescale is high performance computing built for the cloud, to empower engineers while giving IT security and control. From supersonic jets to personalized medicine, industry leaders are bringing new product innovations to market with unprecedented speed and efficiency with Rescale – the cloud platform that delivers intelligent full-stack automation and performance optimization. IT leaders use Rescale to deliver HPC-as-a-Service with a secure control plane to deliver any application, on any architecture, at any scale on their cloud of choice.