Case StudyTransforming semiconductor EDA into an enterprise SaaS
Design focused on opening a new market
EDA is a class of enterprise software used by engineers to design the chips found in cars, phones, computers, and almost everything else used in our daily lives. Commercial EDA tools have traditionally been expensive, feature-rich, highly configurable, and designed by hardware engineers, for hardware engineers. From a UX point of view, existing EDA software has always been highly complex, with each engineer requiring days or weeks of vendor training. Complexity and cost have been barriers to entry for many companies, especially ones without a large enough business case to justify the expenditure. These underserved companies must find alternatives to designing their own custom ASICs (e.g. using more expensive FPGAs).
OpenROAD, an industry-funded project out of the University of California-San Diego VLSI Laboratory, focuses on these businesses by building an EDA solution that processes RTL to GDSII without human-intervention, run on a single OS instance. Aligned to their mandate to democratize system and product innovation in silicon, Planorama started with the installed solution and transformed it into a cloud-based SaaS that enables businesses to leverage this EDA solution without installation or maintenance, and with comparably very little training requirements. With our solid understanding of both existing commercial landscape of EDA solutions and modern enterprise SaaS architecture, Planorama designed OpenROAD Cloud to be accessible by non-chip design engineers while enabling them to easily configure constraints for the project’s cloud-compute jobs with specific power, performance, and area goals.
Accessible from anywhere
Rethinking EDA from the ground up as a SaaS solution, we understood that the traditional constraints of installed EDA tooling need not carry over to the cloud. Hardware designers may not always be in front of their computer, and yet want to view the status of their project, and even manage the cloud jobs runs while on the go.
Our design philosophy for OpenROAD Cloud began with a fully-responsive user interface equally accessible by users on desktops, laptops, tablets and mobile phones. While securely logged into their company’s private tenant, users can perform many of the same functions on their mobile phone as are available on their desktop.
Architected to scale in a public or private cloud
As a cloud-native solution, from the beginning we architected OpenROAD Cloud to take advantage of the economies of scale and flexibility of public cloud infrastructure, while providing the option to deploy in a private cloud for companies with data security or compliance concerns. The platform can scale up or down elastically on demand so customers only pay for the compute resources they use while running jobs. Leveraging modern SaaS approaches such as containerization and microservices minimize the footprint and operational overhead of the solution.
Using such an architecture, OpenROAD Cloud was designed to abstract away underlying infrastructure from users, and instead focus on their goals and workflows, including the achieving certain power, performance, and area (PPA) objectives for their hardware designs. We were also able to leverage recent Ph.D. research performed on the original OpenROAD application to support auto-tuning: a method of running many cloud jobs in parallel with algorithmically-honed constraints that identifies the best constraint set that achieves the user’s PPA goals.
By partnering with Planorama, OpenROAD Cloud users will be able to access a powerful EDA tool from anywhere, with low training requirements and no need for costly hardware or maintenance. Our cloud-native architecture enables OpenROAD Cloud to scale up and down with ease, and abstract away underlying infrastructure for users. Plus, with the ability to tune constraints for their designs, OpenROAD Cloud provides users with a premium level of control and optimization for their chips, providing an alternative to expensive FPGAs by enabling previously underserved companies to design custom ASICs.