Looking to Roll Out Redesigned Rubber

The Goodyear Tire & Rubber Company opened in 1898 as a manufacturer of tires for bicycles and carriages. In fewer than 20 years, the company grew into the world’s largest tire company. In large part, this expansion emerged from ongoing innovation.

“Innovation and technology have been a part of the DNA of Goodyear tires, services and technology for more than 125 years,” says Chris Helsel, the company’s senior vice president of global operations and chief technology officer. “Throughout our history we have always looked for ways to evolve the way we work to improve efficiency and performance.”

Over the decades, computer technology has played an increasing role in Goodyear’s evolution. In 1968, for example, Goodyear Research installed a data-acquisition system that, according to the company’s history page, made its “laboratories the most highly computerized in the rubber industry.”

Today, Goodyear makes use of high-performance computing (HPC), which Helsel calls “an essential part of Goodyear’s technology strategy, especially the development of computational models that simulate different tire performances. These capabilities are a crucial part of Goodyear’s product-development process.”

Those capabilities expanded even more as Goodyear’s scientists worked with the Department of Energy (DOE) Exascale Computing Project (ECP) over the past several years as a member of the ECP’s Industry and Agency Advisory Council.

Chris Helsel
Senior Vice President of Global Operations and Chief Technology Officer,
The Goodyearr Tire & Rubber Company

Exascale Opportunities

“The next generation of HPC—exascale—is an important development for the industry and Goodyear,” Helsel says. “The availability of exascale computing will provide greater opportunity to discover new materials, enable deeper examination of complex designs and create new approaches to optimize business systems.”

Goodyear already puts HPC to work in the design stages. “Before attempting to build a new tire prototype, Goodyear engineers use HPC to actively explore the tire design universe,” Helsel explains. “Exploration efforts for a single product can include running up to 300 different models, using 1,000 to 10,000 CPU processors simultaneously.”

As a start toward even faster computing, Goodyear is building a network of experts. As Helsel puts it, “Our interaction with the ECP has presented a valuable opportunity to interact with HPC developers, allowing Goodyear to establish connections with programmers, manufacturers, other industry adopters and national agencies.” Here, Helsel reflects on one of ECP’s key objectives: combining today’s top HPC and experienced computational and computer scientists to spawn the development and application of technology and tools that drive engineering campaigns at previously impossible scales.

Those networks are already benefiting Goodyear. For one thing, Helsel notes that being part of the ECP “has enabled us to be early adopters, and we have been able to stay informed about the emerging shifts in HPC architecture,” especially the move from systems based on central processing units (CPUs) to hybrids that use CPUs and graphic processing units (GPUs). ECP provides crucial support for CPU/GPU hybrid computing. As one example, ECP developed many unique software products, including exascale-ready applications for advancing material design. Goodyear and other companies and institutions can make use of these products to transform existing software to run on today’s accelerated platforms.

The company also plans to dig even deeper into tire designs. “With the advancement of Goodyear’s HPC infrastructure towards exascale, an aspect that will be explored to a much broader extent will be the impact of variability of both material characteristics and design elements on product performance,” Helsel says. “At the moment, this is a rather daunting task due to the sheer number of variations that need to be explored simultaneously, but addressing this challenge will help Goodyear design new products more efficiently.” Exascale capability and the impact on HPC will turn daunting into doable.

Creating New Code

Before running simulations on exascale systems, though, Goodyear must adapt its existing software. For now, the company is using the Kokkos C++ Performance Portability EcoSystem, which is part of ECP. “Kokkos provides a performance-portable programming model,” says Michael Heroux, a senior scientist at Sandia National Laboratories and ECP’s director of software technology. “This means it allows code to be written in a way that can run efficiently on various types of hardware architectures, including those that use AMD GPUs,” such as ORNL’s Frontier and El Capitan at Lawrence Livermore National Laboratory.

Revising software to run on GPUs provides many benefits, and the key one is enhanced performance. As Heroux says, “Transitioning to GPU-capable codes means leveraging the parallel processing capabilities of GPUs, which is essential for success since the vast majority of the performance of these systems comes from the GPUs.” Nonetheless, making code ready for accelerated platforms is complicated. As some of the key challenges, Heroux points out “the need for new programming models and algorithms, and addressing these challenges was an essential focus of ECP.”

In the future, Goodyear will also use ECP’s open-source Extreme-scale Scientific Software Stack (E4S). Working with source code developed with a range of HPC products, including various math libraries and programming models, E4S creates code that can run on virtually any HPC platform, including GPU-enabled architectures. Even more, E4S provides some future-proofing to the exascale transition. “The standardization of the new software stack also enables better collaboration and future sustainability of proprietary software systems,” Helsel says.

In addition to exascale-related tools, Goodyear gains even more from ECP, Helsel says. “Being a part of the ECP gives Goodyear’s engineers an opportunity to learn about the software-development paradigm shift by directly interacting with scientists and engineers from different national labs.”

For Goodyear, though, its exascale opportunities are only beginning. “Goodyear is planning on transitioning our HPC infrastructure toward exascale over the next three to five years,” Helsel says. “Besides exploring the impact of variability on our product performance, we are looking to develop enhanced business planning and scheduling models.” As one example, he says, Goodyear is “looking forward to introducing stochastics aspects into our business modeling.” That business modeling could also include the incorporation of risk management, “rather than seeing it as an additional analysis step.”

Overall, Goodyear plans to take advantage of as much of the learning they have gained from their relationship with ECP as possible. Already, Helsel says, “Working with ECP has allowed Goodyear to learn more about the potential opportunities that could be beneficial to the organization and how to best take advantage of the new exascale infrastructure.”