Reflecting on NVIDIA’s AI-powered standard cell verification methodology
Interested in NVIDIA’s presentation at User2User 2024 Conference? Watch the recording here.
The recent User2User North America conference proved to be a pivotal gathering for the IC design community. Siemens EDA facilitated an environment for industry professionals to exchange experiences and insights, fostering learning and growth.
One standout presentation came from Chengcheng Liu, an Engineering Manager at NVIDIA, focusing on “Utilizing Solido Additive Learning Technology to Speed Up Standard Cell Library Verification”. This presentation highlighted NVIDIA’s AI-powered methodology for the standard cell library verification including the use of Solido’s Additive Learning technology that allowed them to save across incremental and iterative runs.
NVIDIA’s approach, which emphasizes the use of Artificial Intelligence to optimize the verification process, represents a shift towards more efficient and accurate design verification. The Additive Learning technology offers a significant boost to existing AI techniques, providing speedups on incremental verification runs. This is crucial as the industry moves beyond the 5nm node, where design complexity and verification challenges significantly increase.
For those interested in Custom IC Verification sessions, I recommend exploring the User2User conference website and watching the recordings.
