Check out the on-demand sessions from the Low-Code/No-Code Summit to learn how to successfully innovate and achieve efficiency by upskilling and scaling citizen developers. Watch now.


Nvidia and Rescale today announced several enhancements designed  to simplify artificial intelligence (AI) development and optimize high-performance computing (HPC) workflows. Nvidia is powering a new AI compute recommendation engine (CRE) to replace a more manually tuned approach. It’s also integrating the Nvidia AI platform into Rescale’s HPC-as-a-service offering. 

Both developments  promise to make it easier to spin up new scientific workloads and operate them more efficiently. This will also apply equally to public cloud service and private cloud infrastructure. 

Rescale specializes in tools for automating scientific computing workloads — a  field that is ripe for disruption, since engineers may sometimes spend more time configuring experiments than running them. Earlier this year, Rescale announced tools to help refactor legacy apps to run on containers to dramatically simplify configuration and deployment. 

It also announced a partnership with Nvidia in July to containerize many Nvidia workloads. The latest news builds on this partnership to automate support for Nvidia’s AI platform. This will automates the use of AI for physics, recommendation engines, simulations, medical research and more. It also applies Nvidia’s recommendation capabilities back on the HPC infrastructure itself. 

Event

Intelligent Security Summit

Learn the critical role of AI & ML in cybersecurity and industry specific case studies on December 8. Register for your free pass today.

Register Now

AI-powered infrastructure recommendations

Spinning up scientific computing workloads requires a delicate balance involving hardware, networking, memory, software and specific configurations. Rescale and Nvidia have collaborated on what the two  are billing as the world’s first AI-powered recommendation system for HPC and AI workloads. The companies claim it will  assist teams with balancing decisions about architectures, geographic regions, price, compliance and sustainability objectives. Nvidia and Rescale trained the system using data from more than 100 million production HPC workloads. 

“Prior to compute recommendation engines, the primary way we provided compute optimization was through our solution architects working with the customers guided by our internal benchmarks library,” Edward Hsu, Rescale’s chief product officer told VentureBeat.  “With the compute recommendation engine, we are bringing unprecedented levels of automation and insights by applying machine learning [ML] to infrastructure telemetry and job performance data.”

With the new engine, users choose a workload and Rescale will suggest a computing architecture to provide the best performance. Hsu claims that these recommendations are 90% accurate. Further optimization will also need to account for the models, which can impact both performance and the applications they run on. 

Rescale is also integrating the Nvidia Base Command Platform software to orchestrate workloads across clouds and on-premises Nvidia DGX systems.

Expanding the reach and utility of AI

The two companies are also partnering to support the Nvidia AI Enterprise Software Suite on top of the Rescale platform. Soon, this will help automate workflows using tools like Isaac for programming robots, Nemo for languages, Merlin for recommendations, Morpheus for Security and Holoscan for medical AI. Nvidia Modulus, a physics-ML framework, is also now available on Rescale — which will play a key role in helping companies create faster digital twins for simulating the physical properties of products and equipment.

Existing AI frameworks on Rescale, such as PyTorch and TensorFlow, are more general purpose. Modulus is a programmable physics-informed neural network that can create models that the company claims run hundreds or thousands of times faster than traditional simulation techniques. The Modulus support allows teams to more easily apply AI to emulate physics-based simulations at much higher performance and lower cost  

“As we see the engineers move from intuition-based engineering towards AI-assisted engineering, bringing together the tools for computational engineering and artificial intelligence will be critical to help companies accelerate new product innovation,” Hsu said.

VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.