Delivering Universally Optimized IP with RPE
Introduction
For IP providers, power optimization is a unique challenge. Unlike SoC designers, they don’t know exactly how or where their IP will be used.
An IP core might end up in a mobile SoC, a network switch, or a data center accelerator — each with very.
The mission: Deliver “always good” power savings — optimizations that work everywhere, without breaking functionality in any possible integration.
Challenge
Optimizing IP for Unknown Applications
- Diverse Use Cases: The same IP must run efficiently in countless SoC architectures.
- No Specific Stimulus: Customer-specific simulation vectors aren’t available at IP development time.
- Universal Safety: Any optimization must be safe in all contexts — no risky cross-boundary logic or integration-specific gating.
Zettabolt Solution
RPE for High-Quality, Localized Gating, The IP provider turned to Zettabolt’s RTL Power Explorer (RPE), configuring it for universal power savings:
- Vectorless Mode: Analysis without requiring stimulus data, ensuring opportunities are not tied to a single use case.
- Localized Optimization: Restricted RPE to clock-gating within each design hierarchy, preventing any optimization that could fail in a different integration.
- High-Impact Simplicity: Focused on finding clean, high-quality enables that are simple to implement and review.
Results
Fast, High-Impact, "Always Good" Savings
The tailored RPE flow delivered exactly what the IP provider needed:
- Runtime: < 2 hours on designs ranging from 5–10 million gates.
- Impact: Found high-quality clock-gating enables for 15% of registers.
- Ease of Adoption: All identified changes were simple to verify and integrate, making them universally safe.
With RPE, the provider shipped a more power-efficient IP core, instantly increasing its value and market appeal across a broad range of customers.