EDA companies (Electronic Design Automation) play a vital role in the semiconductor industry. They provide the tools to design and virtually test chips that power electronic systems such as smartphones, cars, and computers. But the challenges this industry faces can be considerable, with complex hybrid cloud management requirements.
Designing these chips is no easy feat. It requires thousands of computers and users to collaborate over vast and intricate data sets and files that must all be accessible and in-sync. Technological advancements in fields such as IoT, AI, 5G, and automotive have given rise to even more complex chip designs, putting additional pressure on EDA corporations. Many EDA companies are turning to the cloud to expand their data storage capacity, but this move comes with its own string of challenges.
In this blog post we explore some of the major EDA cloud storage challenges and how companies can address them with the help of NetApp’s Cloud Volumes ONTAP.
The EDA market is projected to reach $9.49 Billion by 2024. Its design tools are critical to semiconductor operations, laying out the map to manufacturing chips reliably and cost-efficiently.
An EDA company must virtually emulate chips, a task that requires extremely large volumes of data. Nearly every step of the way—from design through verification and virtual testing—generates information, and every project creates additional scratch workloads. As a result, the process often involves thousands of computers working together for weeks on enormous data sets and many shared files, typically shared over NFS.
Recent technological developments are rendering this process more complicated. The 5G market, for example, is developing fast, at an expected CAGR of 11.5%. This is driving the demand for 5G-compatible parts and testing tools that can validate them. These parts are faster, more complex, and data intensive.
Similarly, new trends in the Automotive industry such as connectivity, electrification and autonomous driving are also pushing EDA vendors to quickly expand their capacity. These technologies often require the handling of an astonishing amount of data. AI chips for ADAS (advanced driver-assistance systems) can contain 21-25 billion transistors, over 100 billion connected nodes, and 1 trillion pure RC components. On top of this, strict safety regulations laid out by regulatory bodies such as the AEC, require unprecedented levels of virtual chip verification, setting the bar even higher.
EDA companies often struggle to keep up with the flood of new projects that come their way. Their data centers run out of storage space, putting new projects on hold until current projects come to an end. To overcome this problem, companies must find a way to spin up extra computers and storage space quickly. This is virtually impossible when relying solely on in-house infrastructure. Enter the cloud.
Rather than purchase or rent additional storage space, EDA companies are turning to the cloud to quickly gain access to more storage space. This, however, poses several challenges.
Firstly, it’s difficult to know ahead of time what data will be needed for a particular workflow. Chip design is highly complex, programs depend on previous run results, and on other programs, libraries, and data sets. Predicting what data will be needed from these vast and complex datasets can be very time-consuming. In addition, because this task is so intricate, companies end up syncing too much data to the cloud, unnecessarily spending extra time configuring resources that ultimately won’t be used.
Secondly, read-intensive EDA workflows can run into performance challenges: data must be highly available to many different users in disparate locations, all without affecting metrics such as latency.
Thirdly, in the EDA design process, data collaboration is key. Different users access the same datasets, often at the same time. In hybrid architectures, clients and group memberships on-premises both have unique permissions, and all of them must be mirrored in the cloud. In addition, multiple copies of data need to be available across the cloud as well as on-premises, requiring repeatedly syncing data. This is complex and can result in out-of-sync data, locking issues and access problems, which might lead to slower processes and an increased time to market.
Another constraint is that most EDA companies do not own their own FABs, so they must reserve them long before the designs are complete. This means that any delay in design could cost millions of dollars.
Overcoming these cloud challenges and utilizing data as efficiently as possible is critical for EDA companies. Since the design of a single chip costs nearly the same as that of designing a million chips, the ability to complete a design on schedule and with full confidence is crucial.
One way to overcome these storage challenges is to use cloud bursting. In this approach, applications running in a data center or private cloud can burst to the public cloud when extra EDA cloud computing capacity is needed. By bursting to the cloud, when a new and data-heavy project comes in, EDA companies can ensure that the extra storage space and computers are available on demand. However, this still leaves EDA companies with the challenges of data copies, data syncs, and performance. That’s where NetApp Cloud Volumes ONTAP comes in.
Cloud Volumes ONTAP, NetApp’s enterprise-grade storage management solution for AWS, Azure, and Google Cloud facilitates bursting to the cloud while also solving the challenges that come with hybrid environment, data collaboration, and resource allocation.
Cloud Volumes ONTAP makes it incredibly easy to burst to the cloud: users can spin up a Cloud Volumes ONTAP instance, mount it, and give users access all with a few clicks, providing limitless cloud resources. And as soon as the cloud resources aren’t needed, tear it down until you need to burst to the cloud again.
FlexCache® enables quick replication of data on-premises to a target Cloud Volumes ONTAP instance in a remote location. This capability allows users access to the same data no matter where they are, while the source and copy versions are synced constantly.
Using intelligent NVMe caching, Cloud Volumes ONTAP minimizes cloud storage latency, enhancing both user experience and application performance.
These features combine to offer benefits that can significantly optimize and shorten EDAs’ chip emulation lifecycle. With Cloud Volumes ONTAP:
This American multinational corporation creates software, services, and semiconductors in the field of wireless technology. It was facing the challenges that chip designers deal with: constantly running out of storage space, out-of-sync data, long and complicated design processes, and highly complex infrastructure which was difficult to mirror in the cloud.
The company combined Cloud Volumes ONTAP on AWS with FlexCache and intelligent NVMe caching to achieve:
The demand for chip design is growing by the day, but as long as EDA companies remain confined to their legacy environments, they will only be able to handle a limited number of projects at a time.
This is where cloud bursting comes in: a valuable tool which allows companies to scale their storage capacity up and down as needed to accommodate the data-intensive chip design process. However, since there are many collaborators involved, multiple copies of data need to be made available and maintained without affecting performance.
Cloud Volumes ONTAP not only enables EDA companies to easily burst to the cloud, but its replication and caching technologies help overcome major hybrid storage challenges. Using NetApp’s solution, EDA companies can effectively share and work on their numerous data files, taking on more projects at a time, and ultimately accelerate chip design to meet their tight production schedules.