STA analysis for high performance GPU core

STA analysis for high performance GPU core

Sudhanshu Dubey1, Dr. Eleena Mohapatra2

1Sudhanshu Dubey, Department of Electronics and Communication Engineering, RV College Of Engineering

2Dr. Eleena Mohapatra  Department of Electronics and Communication Engineering, RV College Of Engineering

---------------------------------------------------------------------***---------------------------------------------------------------------

Abstract - High-performance GPU cores are essential for handling complex computations and parallel processing tasks, significantly outperforming traditional CPUs. The architectural complexity and diverse operating conditions of GPUs necessitate rigorous design and verification processes, with Static Timing Analysis (STA) being crucial for ensuring performance and reliability standards. This project enhances STA for high-performance GPUs by developing automation scripts using Perl, Python, and Flask, streamlining the process, reducing manual effort, and minimizing errors. Techniques such as VT swapping, buffer insertion, clock pushing, and advanced crosstalk mitigation are employed, demonstrating significant improvements in timing performance. The findings provide valuable insights for engineers and designers, contributing to the advancement of STA practices in the semiconductor industry.

.

Key Words: High-Performance GPU, Static Timing Analysis (STA), Automation, VT Swapping, Buffer Insertion, Clock Pushing, Crosstalk Mitigation, Timing Performance, Semiconductor Design, Digital Circuit Verification