Presentation
A design analytics-based methodology for enhancing Dynamic IR signoff with minimum design changes.
DescriptionIncreasing functionality and design complexity of Automotive ADAS SOCs has resulted in increasing power grid complexity with high Dynamic Voltage Drops creating significant challenges to close IR signoff at lower FinFET nodes. Dynamic Voltage Drop (DVD) leads to change in cell delays that result in degrading timing QoR and performance. The two main challenges that designers constantly grapple with today for efficient DVD signoff are (1) lack of automated flows for debugging the actual root cause of dynamic IR due to which fixing IR violations becomes time consuming and (2) ensuring high IR coverage with peak power/activity-based vector selection for Dynamic IR signoff. Temporal ‘Total' power/activity-based vector selection can miss spatial local PDN issues. We present here a design analytics-based methodology for enhancing Dynamic IR signoff with minimum design changes that addresses both the above challenges. Our results using DVD Diagnostics methodology show significant decrease in the number of IR violations to fix. Using design analytics on long VCDs, we could perform fast temporal and spatial profiling of long vectors to increase IR coverage for VCD based methodologies.
Event Type
Engineering Track Poster
TimeTuesday, July 11th5:00pm - 6:00pm PDT
LocationLevel 2 Exhibit Hall
Back-End Design
Embedded Systems
Front-End Design
IP
RISC-V