Cross-Stage Power Prediction for the VLSI Circuits
A demo.- Advisor: Professor Yibo Lin.
- Proposed an innovative VLSI circuit power prediction framework that integrates cross-stage circuit-transformation-aware learning.
- Achieved a significant reduction in error rates and computation time compared to industry-leading commercial tools, verified using CircuitNet benchmarks.
- It thus enabled effective power optimization strategies in early-stage VLSI design.
Yufan Du
PhD Student
My research focuses on domain-specific accelerators, compilers, LLM serving systems, and ML systems.