High-Quality Silicon With Cloud-Based Verification

New materials, vertically stacked architectures, and angstrom-level process technologies—the complexity of today’s SoCs continues to grow to meet the needs of demanding applications such as AI, autonomous vehicles, and high-performance computing. This trend only places greater pressure on verification, already notorious for being a significant bottleneck in chip development.

Design teams can’t afford to miss any bugs. But how exhaustive can their verification processes truly be in the face of limited compute resources?

Cloud-based verification provides an answer. With its elasticity, scalability, and flexibility, the cloud enables verification engineers to tap into the compute resources they need when they need them. In this article, I’ll highlight the advantages of moving verification to the cloud through the experiences of Imparé, a chip verification services company that’s committed to helping its customers deliver high-quality silicon.

Virtual server farm facilitates verification

With offices in California and Pakistan, Imparé has a team of engineers with more than 70 years of cumulative design and verification experience. Many actively contribute to help verify RISC-V and OpenTitan cores. They have taped out over 40 chip designs for some of the world’s leading global semiconductor companies.

The compute infrastructure needed to run simulations and regressions to verify chips can be massive and costly. Yet, powerful servers are needed to catch corner-case bugs that lurk beneath the layers. For Imparé, rather than spending time and resources to build out a separate server farm for its compute, storage, and license needs, the company has looked to the cloud. Specifically, the company has implemented cloud-based verification via a software-as-a-service (SaaS) model to access cloud-native electronic design automation (EDA) tools.

“Through SaaS cloud-based verification, we got a virtual server farm for Imparé and were able to focus on the verification challenges presented by our customers,” said Faisal Haque, the company’s founder. “During our evaluation, we found the environment easy to set up and get going. CAD/IT is fully supported so any installation complexity was removed. The interface is straight-forward and once set up, it gives you a Linux terminal to work from. Thanks to the quick ramp up, we didn’t waste time and could stay focused on verifying the designs. We also benefit from a flexible licensing scheme, which allows us to easily scale our verification computation needs up or down. And remote access based on an HTML browser interface with multi-factor authentication provides tight security.”

The team’s cloud-based verification system includes a complete framework from integrated development environment to the ability to specify coverage targets and test plans and to automate and build custom scripts. The engineers can create testbenches with SystemVerilog, UVM, VHDL, and mixed language designs.

Preventing bug escapes and strengthening collaboration

With a globally distributed verification team, Imparé needed a verification environment that supports easy collaboration. To meet this requirement, the company selected its cloud-based verification simulation flow for its inclusion of debug and analysis capabilities with pre-configured verification tasks. “With this environment, our team can check out and check in verification projects remotely, saving the work that has already been completed and sharing it with their peers,” said Haque.

Designs that the company’s customers create are fairly complex. Not only would doing verification piecemeal be challenging, it could also lead to errors and bug escapes. This is why a cloud-based solution providing a complete verification flow, spanning test planning to regression with automation built in, was attractive to Haque and his team.

Since Imparé engineers work on several designs in parallel, reaching verification closure quickly and accurately is especially important. Understanding the level of coverage can be challenging given that every tool in the flow has different metrics. Engineers often must spend considerable time interpreting this data manually, an approach that could be hampered by human error. Having to rerun failed tests takes up more time and also can increase the chance of bug escapes. With its cloud-based verification solution from Synopsys, the team can take advantage of automation that can prevent lengthy verification closure and avoid missed bugs.

The team also experiences spikes in its verification work during crucial milestones. Cloud-based verification allows them to quickly scale compute resources up and down as needed during times of peak and waning demand. With this flexibility, the team can also optimize its verification costs. In addition, should the team be interested in further boosting productivity with enhancements such as AI-driven verification, the cloud provides the ideal environment for running such tools.

“Our experience with our Synopsys Cloud verification solution demonstrates its benefits for both startups and large companies,” said Haque. “Many startups would not be using on-prem resources to their fullest, making those resources unproductive and expensive to license and maintain. A startup can use the Synopsys SaaS Cloud solution to build out an infrastructure with a limited and flexible investment model. Larger companies can benefit from the ability to scale as needed and augment their existing on-prem resources to meet peak demands. Startups and large companies alike can take advantage of the remote collaboration opportunities that the Synopsys Cloud can provide.”