From the smartphones in our hands to the big data centers powering the internet, from self-driving cars to robotic surgical systems — inside each is a tiny piece of technology that makes it all possible: a semiconductor. We’re seeing billions being invested in semiconductors across the globe. Intel is spending $20 billion to double down on chip manufacturing, South Korea is joining the global chipmaking race with a $450 billion spending plan, even Bosch, a tier-1 Automotive supplier, is building its own chip fab – these are just a few examples of big investments.
Infrastructure will play a critical role in enabling organizations to maximize return on these investments. Semiconductor design is essentially a high-performance computing (HPC) workload and is massively parallel. As the number of concurrent jobs increases, they often place the underlying infrastructure under tremendous stress, forcing unprecedented growth in essential IT resources such as compute, networking, and storage. A decade ago, CPU performance was a barrier, and the industry relied more on concurrent processes and CPUs with more cores. As total cores grew, the bottleneck shifted to storage.
A semiconductor design project that previously required hundreds of terabytes now requires petabytes of data. This data growth demands nothing less than a purpose-built HPC infrastructure solution that can support hundreds of thousands to millions of jobs. At Dell Technologies, we’re committed to advancing HPC and our HPC experts are active innovators and collaborators in the worldwide technical community. We provide scalable, flexible HPC solutions designed to help you solve complex problems faster than ever. Our portfolio for HPC infrastructure spans workstations, servers, networking, storage, rack systems, and services. We deliver the compute, throughput, and capacity needed to manage the rapid data growth and increased workload demands. For example, here is a Dell Ready Solutions for HPC Digital Manufacturing with Altair HyperWorks.
Dell Technologies has set a new standard for unlocking the potential of semiconductor design tool performance with Dell PowerScale Storage. The PowerScale family, with its scale-out architecture, is optimized for HPC, providing the performance and efficiency needed to take on the challenges of supporting extreme workloads. The PowerScale all-flash platform – powered by the PowerScale OneFS operating system – provides a powerful scale-out storage architecture for fast access to data to speed up verification and regression tests to meet time-to-market pressures.
The job scheduler is a key infrastructure component of virtually every HPC environment – including all semiconductor design tool workflows. Altair Accelerator is a high-performance enterprise-grade job scheduler designed for distributed HPC environments. Integration of Dell PowerScale with Altair Accelerator enables Storage-Aware Grid Acceleration (SAGA), an elegant quality of service (QoS) job scheduler innovation that can manage compute, networking, and storage holistically – scheduling semiconductor design jobs to maximize throughput at the data center level.
Emerging technologies like artificial intelligence (AI) and real‑time data analytics are creating exciting opportunities in the semiconductor industry. These workloads are typically fed by massive amounts of data and are ideally suited to run on HPC systems. PowerScale is the ideal storage complement to server GPU-accelerated compute for AI workloads. PowerScale effectively compresses the time needed for training and testing analytical models for multi-petabyte data sets. PowerScale enables semiconductor companies to process more training data in less time, ultimately improving algorithm development and manufacturing quality as well as predictability.
To find out more on how your organization can take advantage of Dell PowerScale to achieve optimal performance for HPC workload in semiconductor design environments, visit our website.
