July 2022 Summaries

Rescale, a leader in high-performance computing (HPC) for the cloud, is collaborating with NVIDIA and Microsoft to enhance HPC and artificial intelligence (AI) solutions, aiming to advance engineering and scientific innovation by improving research and development efficiency. This partnership, building on Rescale's previous collaborations with Microsoft and NVIDIA, offers a comprehensive platform that integrates software and hardware to accelerate innovation in various industries. Rescale simplifies the deployment of GPU-optimized applications using NVIDIA NGC and Microsoft Azure, enabling rapid scaling and execution of AI and HPC workflows. By providing access to NVIDIA GPU-based instance families on Azure, the collaboration addresses the growing demand for fast and efficient AI-driven HPC applications, helping industries like aerospace, automotive, energy, and life sciences reduce simulation and training times. This initiative will also include a series of upcoming content, such as webinars and demos, to demonstrate the ease of adopting these advanced technologies for digital R&D.

Scientific and engineering design has evolved from traditional prototyping to the use of simulation, and now increasingly incorporates artificial intelligence (AI) to enhance research and development (R&D) processes. The integration of AI into simulation allows for more efficient exploration of design spaces, reducing the computational demands traditionally associated with physics-based simulations. While AI offers transformative potential in fields ranging from aerospace to biotechnology, its adoption presents new IT challenges as organizations must integrate AI tools with conventional solvers to create cohesive workflows. Rescale, in collaboration with NVIDIA, addresses these challenges by offering NVIDIA Modulus on its platform, which streamlines the development and deployment of AI-driven simulations. This platform facilitates real-time inference using AI surrogate models, significantly reducing design time and enhancing productivity. By harnessing the power of NVIDIA's latest GPU technology, Rescale provides a comprehensive, cloud-based solution that simplifies the implementation of AI in R&D, enabling faster and more innovative product development.

As high-performance computing (HPC) simulations grow increasingly complex, optimizing compute resources for these tasks has become a significant challenge due to the vast array of available hardware and software options, especially in cloud environments. Unlike traditional on-premises HPC setups, which are limited and quickly outdated, cloud-based HPC continuously benefits from the latest advancements, making it crucial to select the best configurations for performance and cost-effectiveness. Artificial intelligence (AI) is emerging as a valuable tool in this context, leveraging extensive metadata from past simulations to recommend optimal hardware-software combinations, thereby enhancing workflow efficiency without the need for users to stay updated on every new chipset release. Rescale, a cloud-based platform, exemplifies this approach by using AI to automate HPC workload management across multiple cloud providers, allowing engineers to focus on their core tasks while benefiting from AI-powered suggestions that enhance throughput and reduce downtime. This AI-driven optimization is increasingly necessary as the market diversifies, with new technologies continuously emerging and creating new opportunities for performance gains in critical applications such as motorsport simulations.

Rescale has expanded its collaboration with Google Cloud by integrating new Arm-based Tau T2A virtual machines into its Google Compute Engine (GCE) and Google Kubernetes Engine (GKE) offerings, under the Coretype name "Trona." These T2A instances, built on the Arm architecture and featuring Ampere Altra processors, provide significant cost-performance benefits for high-performance computing (HPC) and high-throughput computing (HTC) workloads, particularly in AI and electronic design automation (EDA) applications. This development allows Rescale's customers to efficiently leverage these resources through a user-friendly interface, facilitating rapid deployment and innovation in scientific and engineering simulations. The growing ecosystem of Arm-supported simulation applications, including tools from Cadence, OpenFOAM, and GROMACS, further enhances the value proposition of T2A, offering customers a robust library of applications that capitalize on the sustainability and performance advantages of the Tau T2A instances.

R&D and engineering teams increasingly require high-performance computing (HPC) resources to manage complex, computation-intensive applications. With the rise of specialized HPC hardware options in the cloud, organizations face new challenges in optimizing their HPC operations to enhance efficiency and innovation. This guide explores how IT and HPC leaders can use data-driven benchmarking to capture more value from the latest cloud-HPC technologies, focusing on optimizing hardware and software configurations for specific business goals. By leveraging a framework that includes performance intelligence and continuous benchmarking, organizations can improve workload performance and reduce costs, as demonstrated by companies like Vertical Aerospace and Arrival. The guide emphasizes the importance of adapting quickly to new technologies and maintaining a flexible, multi-cloud strategy to achieve significant cost-performance gains and ensure the success of large-scale projects.

Engineers are increasingly turning to cloud-based high-performance computing (HPC) to enhance multidisciplinary design optimization (MDO) processes by integrating various design disciplines, as opposed to traditional on-premise systems, which are often limited by capacity and cost issues. Modern aircraft designs, such as the Airbus A350 and Boeing 787, exemplify the benefits of MDO by optimizing across aerodynamics, structural analysis, and other fields to achieve superior designs that improve fuel efficiency and reduce emissions. However, on-prem HPC systems struggle with the growing demand for complex simulations, leading to long queue times and inefficiencies. Cloud solutions offered by providers like AWS, Azure, and Rescale alleviate these challenges by offering scalable resources, enabling engineers to conduct high-fidelity simulations rapidly and cost-effectively under an OPEX model. This shift not only democratizes access to advanced computational resources but also supports new entrants in the engineering sector by removing the capital investment barrier. Additionally, Rescale's platform offers comprehensive cost visibility and AI-driven recommendations to optimize resource use, ultimately enhancing the capacity for innovative design exploration and leading to improved engineering outcomes.