
Nvidia will Improve AI Cloud Automation and HPC as a Service in Collaboration with Rescale
Nvidia (NASDAQ: NVDA) and Rescale unveiled several improvements aimed at streamlining the creation of artificial intelligence (AI) and streamlining workflows for high-performance computing (HPC).
By
CIO Applications Europe | Friday, November 18, 2022

Nvidia is powering a new AI compute recommendation engine (CRE) to replace a more manually tuned approach.
FREMONT, CA:Nvidia (NASDAQ: NVDA) and Rescale unveiled several improvements aimed at streamlining the creation of artificial intelligence (AI) and streamlining workflows for high-performance computing (HPC). Nvidia is powering a new AI compute recommendation engine (CRE) to replace a more manually tailored method. Additionally, the Nvidia AI technology is being incorporated into Rescale's HPC as a Service offering.
The ability to quickly spin up new research workloads and run them more effectively is promised by both breakthroughs. Both public cloud services and private cloud infrastructure will be covered by this. Engineers may spend more time configuring experiments than actually performing them, hence Rescale specialises in technologies for automating scientific computing workloads. Rescale introduced tools earlier this year to aid in refactoring legacy apps to operate on containers, greatly simplifying configuration and deployment.
Stay ahead of the industry with exclusive feature stories on the top companies, expert insights and the latest news delivered straight to your inbox. Subscribe today.
It also disclosed a collaboration with Nvidia to containerised certain Nvidia workloads. The most recent information strengthens this alliance by automating support for Nvidia's AI platform. This will automate the application of artificial intelligence in physics, recommendation systems, simulations, and more.
Additionally, it applies Nvidia's capability for making recommendations to the HPC infrastructure itself.
AI-powered Infrastructure Recommendations
It takes a careful balancing of hardware, networking, memory, software, and particular configurations to spin up scientific computing workloads. The world's first recommendation system for HPC and AI workloads has been developed in partnership between Rescale and Nvidia. The companies assert that it will help teams balance choices regarding architectures, geographic locations, costs, regulatory requirements, and sustainability objectives. Utilising information from more than 100 million production HPC workloads, Nvidia and Rescale trained the system.
According to the chief product officer at Rescale, before computing recommendation engines, the primary way to provide compute optimisation was through our solution architects working with the customers under the direction of the internal benchmarks library. By integrating machine learning [ML] into infrastructure telemetry and task performance data, we are providing unparalleled levels of automation and insights with the compute recommendation engine.
Users select a workload with the new engine, and Rescale suggests the optimum computer architecture gets the greatest performance. The models, which have the potential to affect both performance and the applications they run on, will also need to be taken into account in further optimisation.
To orchestrate workloads across clouds and on-premises Nvidia DGX systems, Rescale is also integrating the Nvidia Base Command Platform software.
Expanding the Reach and Utility of AI
The Nvidia AI Enterprise Software Suite will be supported by the two businesses on top of the Rescale platform. Using tools like Isaac for programming robots, Nemo for languages, Merlin for recommendations, Morpheus for security, and Holoscan for medical AI, will soon assist in automating workflows. Also available on Rescale is Nvidia Modulus, a physics-ML framework that will be crucial in assisting businesses in developing quicker digital twins for mimicking the physical characteristics of goods and machinery.
More all-purpose AI frameworks exist on Rescale, including PyTorch and TensorFlow. The company says that Modulus, a programmable physics-informed neural network, can generate models that run hundreds or thousands of times quicker than those produced by conventional simulation methods. The Modulus support makes it easier for teams to use AI to simulate physics at considerably higher performance and less expense.
Bringing together the tools for computational engineering and artificial intelligence will be crucial to help organisations accelerate new product invention, as companies observe engineers transitioning from intuition-based engineering to AI-assisted engineering.
More in News
Weekly Brief
I agree We use cookies on this website to enhance your user experience. By clicking any link on this page you are giving your consent for us to set cookies. More info
Be first to read the latest tech news, Industry Leader's Insights, and CIO interviews of medium and large enterprises exclusively from CIO Applications Europe
THANK YOU FOR SUBSCRIBING


