Chip manufacturer NXP Semiconductors, a company counting 29,000 employees in more than 30 countries with revenues of $8.61 billion in 2020, has chosen Amazon Web Services (AWS) as its preferred cloud provider. The company is now migrating the vast bulk of its electronic design automation (EDA) workloads from NXP data centers to AWS.
The Netherlands-based chip developer and manufacturer will leverage AWS’s global infrastructure and capabilities in high performance computing (HPC), storage, analytics, and machine learning (ML) to improve collaboration and electronic design automation throughput across dozens of its global design centers. The company also intends reduce costs through elastic scaling of compute resources and minimize design project scheduling risks. Furthermore, because of AWS’s nearly limitless scalability, NXP engineers would have more time to focus on innovation rather than managing computing resources.
“We believe cloud-based EDA is critical to accelerating semiconductor innovation and getting new designs to market faster to power an increasingly digital world where more and more devices and infrastructure are connected,” said Olli Hyyppa, CIO and senior vice president, NXP Semiconductors. “AWS gives us the best scale, global presence, and selection of compute and storage options, with continuous improvements in price performance, that we need. We’re excited to expand our relationship with AWS to power the next generation of EDA workloads in the cloud. This will give precious time back to our design engineers to focus on innovation and lead the transformation of the semiconductor industry.”
AWS Analytics and ML Services
NXP’s AWS-based operations seek to deliver long-term process gains aimed at revolutionizing how semiconductors are developed and tested. Before NXP can produce new chips, its designs are subjected to intensive testing and validation as part of the EDA process to guarantee they are functionally safe, secure, of high quality, and highly performant. Front-end design, performance simulation, and verification are all part of NXP’s complex EDA processes, as are backend workloads such as timing and power analysis, design rule checks, and other applications used to prepare a device for production.
Traditionally, semiconductor firms have operated these highly repetitive operations from on-premises data centers with fixed computational resources. However, because of the huge computing power required for each cycle, as well as the rising complexity of chip designs, developing a new device might take months or even years unless firms properly predict and build more computational equipment. In contrast, by leveraging AWS to power its EDA, NXP gets the scalability and agility to progress numerous projects at the same time on demand, regardless of complexity, and to conduct dozens of performance simulations in parallel to expedite time to result.
NXP uses AWS analytics and machine learning services to continually optimize its research and development workflows in order to better manage the volume and complexity of its design operations. NXP uses Amazon QuickSight (AWS’s cloud-based machine learning-powered business intelligence tool) to create more sophisticated engineering and operational insights that aid in workflow efficiency. NXP, for example, can minimize the time necessary to repeat semiconductor designs by swiftly converting results from one phase of testing into adjustments for another.
Machine Learning Models, HPC
NXP also utilizes Amazon SageMaker (AWS’s tool that assists developers and data scientists in rapidly building, training, and deploying machine learning models in the cloud and at the edge) to optimize how it arranges computing, storage, and third-party software application licensing. To support this effort, NXP is constructing a data lake on AWS utilizing Amazon Simple Storage Tool (Amazon S3) and AWS Glue (AWS’s service for easily and cost-effectively extracting, manipulating, and loading data).
Furthermore, NXP uses AWS’s portfolio of specialized instances for HPC to further simplify its EDA operations. NXP’s choice of instance types enables it to satisfy the specific requirements of each design project while maintaining a high level of pricing performance. NXP stores petabytes of design simulation data in Amazon FSx for Lustre (AWS’s service that delivers cost-effective, high-performance, scalable storage for computational workloads like EDA) and makes it instantly available for analysis.
“At AWS, we consider ourselves to be a community of builders, and this engagement with NXP reinforces what’s possible when you free builders to work in the best environments, with the infrastructure and capabilities they need,” said Dave Brown, vice president of Amazon Elastic Compute Cloud, Amazon Web Services. “By shifting their EDA workloads to AWS, NXP designers will have access to the best tools available for collaborating on semiconductor design and development around the world. This move will help NXP produce chips that power innovation in IoT, connected cars, and more. We’re proud to support a leading driver of innovation in the semiconductor industry, and we look forward to seeing what becomes possible when chip design moves to the cloud at such a large scale.”