Big Compute 22: Convergence of Cloud, HPC, and AI Creates New Possibilities in Engineering and IT Operations
Yesterday the Big Compute 22 virtual event gathered the world’s pioneers in computing-driven innovation including both the technology providers and the practitioners solving the biggest challenges of our time. Rescale and other participants took the stage to showcase new features, capabilities, and industry advancements that are accelerating the pace of innovation.
What is Big Compute All About?
For a short primer on Big Compute, the community exists to celebrate and discuss the work being done to improve our lives with high performance computing (HPC) and AI/ML tools. Industry practitioners have largely overcome the problems of Big Data and now the bottleneck for many discoveries is Big Compute. From curing disease to launching rockets, scientists and engineers are using unprecedented computing power. Specialized architectures are making this possible in the post Moore’s Law world, but the diversity of computing architectures to optimize workload performance makes this a non-trivial challenge.Big Compute 22 brought together the top minds in HPC hardware, simulation/modeling, and AI/ML to discuss why the industry is shifting to a cloud operating model to improve both IT and engineering outcomes. Notable participants this year included NVIDIA, Microsoft, Intel, AMD, AWS, Ansys, Siemens, NASA, Virgin Orbit, the US Department of Defense, and many others.
Major Themes Throughout the BC22 Sessions
The significant interest in participating in the event from a wide variety of companies made for a jam-packed and exciting agenda. Key event topics across the 20+ sessions provide insight into the trends driving real change for computing practitioners. Some of the major recurring themes include:
- Cloud operating models for HPC quickly becoming the norm for agile digital R&D
- AI/ML-enabled design and engineering accelerates new product development
- The explosion of specialized hardware delivers improvements in performance, cost, and power efficiency
- Computing sustainability and efficiency is something organizations want to control
- Co-innovation and collaboration on new computing systems and products is now commonplace
- Software and algorithms advancement is pushing architecture development to keep pace
Practitioners Leverage Cloud to Operationalize Heterogeneous HPC
Between Big Compute 20 and Big Compute 22 the world shifted dramatically in many ways, especially how and where we work. Cloud-based solutions enabled R&D leaders to continue compute-driven innovation across globally distributed teams and broad application portfolios. Apart from the changes brought about by the pandemic, the dynamics of computing hardware and software have also shifted. One of the most noticeable changes is the fragmentation of computing architectures and the specialization of hardware selection based on the workload. As Moore’s Law has decelerated, gains from heterogeneous architectures like new GPUs, FPGAs, TPUs, ACISs, etc. make cloud (and the flexibility it offers) a much more attractive model than fixed, on-premises deployments.
Rescale Announces Suite of Intelligent Computing Features
Big Compute 22 was also an exciting opportunity to announce several game-changing products and capabilities designed to help customers succeed in their digital transformations. Thriving in this new world of fragmented clouds, architectures, and applications is a challenge for many organizations, which is why Rescale’s CEO, Joris Poort, took this opportunity to announce new features and roadmap for Intelligent Computing. In the world of high performance computing, maximizing performance really matters. So after years of running benchmarks and understanding the factors that slow and speed up HPC jobs, we’ve built the world’s first Compute Recommendation Engine. Now Rescale can not only offer customers access to the latest computing technologies, but we can offer powerful recommendations on how to configure their compute environment to get the results they need.
Rescale Performance Optimization
Rescale’s Compute Recommendation Engine unlocks powerful capabilities for customers, the first of which is performance optimization. Rescale’s Intelligent Computing framework combines AI algorithms and a wealth of computing telemetry data to predict bottlenecks and make recommendations that deliver performance gains over 300% and cost reductions over 500%. With a highly accurate recommendation engine, customers can make data-driven decisions to increase performance across each unique workload. Rescale’s Chief Product Officer, Edward Hsu, shared a few new products that allow customers to incorporate this intelligence into their computing operations: Rescale Compute Optimizer, Performance Profiles, and Continuous Performance Optimization.
Rescale Resource Automation
The second key capability enabled by Intelligent Computing is Resource Automation, the monitoring and control of the various inputs and factors in computing operations. In order to tame the influx of datapoints, Rescale has made significant investments in our control plane that enable full control and automation of HPC operations. Taking advantage of all available clouds, regions, and architectures requires first having a foundation of real-time intelligence. Once this data can be visualized, organizations can then set policies that optimize for a number of business objectives. Equipped with these unique tools, HPC managers can precisely negotiate trade offs between cost, performance, maturity, and even sustainability. This level of fine-grained visibility and control has huge potential for value creation and risk mitigation.
AI-Assisted Engineering and Accelerated Computing from NVIDIA on Rescale
Much like how digital simulation and modeling made huge improvements in engineering and testing, AI/ML-related hardware and software from NVIDIA are transforming how engineers design new products in record time. NVIDIA CEO, Jensen Huang presented the 3 major trends in industrial scientific computing:
- The rise of accelerated computing: Developing new ways to accelerate workload performance despite the end of Moore’s Law computing gains.
- AI turbocharging science and engineering: Merging traditional digital engineering methods with new AI/ML models to explore larger design spaces at greater speeds than ever before.
- The growing adoption of cloud computing: With NVIDIA in every cloud, researchers and engineers can instantly access elastic GPU-accelerated computing from their browser.
The ongoing collaboration between Rescale and NVIDIA has capitalized on these trends and culminated in a big announcement at Big Compute: NVIDIA AI Enterprise will be available to Rescale customers to develop and run their AI applications on Rescale. This partnership brings full-stack HPC and full-stack AI together on Rescale and we are excited to see what customers are able to accomplish. Some impressive examples of these tools in action include NASA’s 4x acceleration of their custom CFD code on Rescale with 5x better cost efficiency. With many new offerings targeting the needs of both IT and R&D pros, NVIDIA on Rescale sets a new standard for automated and accelerated HPC-as-a-Service. Read more about the news from Rescale and NVIDIA in this press release.
Other Highlights from Big Compute 22 You Don’t Want to Miss
The industry advancements revealed throughout the sessions give us a glimpse of how world-class science and engineering teams are driving innovation in their industries.
- Zero emissions vehicles manufacturers like Canoo and Arrival and clean energy computing solutions from Responsible Compute are finding new ways to make both products and computing processes more sustainable.
- Software providers like M-Star are building highly accurate, physics-based models and digital twins to aid the development of biopharmaceuticals.
- Hexagon explored how manufacturing is getting smarter by integrating real-time simulation data.
- Tempoquest shared how they are accelerating weather models run on Rescale and NVIDIA architectures to get more accurate forecasts.
- Convergent Science explains the Simulation Multiverse and how they apply it to evaluate a variety of parameters, scales, and physics to get more accurate optimization results.
- ESI illustrates how the future of digital transformation will include predictive, real-time, and immersive technologies to enable more efficient and collaborative engineering.
- VMware shared how collaboration with Rescale leverages cloud automation to improve the HPC user experience and accelerate performance across multiple clouds.
If you weren’t able to watch the live stream you can still register to access all the sessions on-demand here. In between conferences, BigCompute.org also publishes content on the advancements in HPC, AI, and related technologies so be sure to check back in regularly to follow the action.
If you have questions about any of the above-mentioned product announcements, you can get in touch with us at firstname.lastname@example.org.