Evolution of Silicon Design

Silicon design is undergoing a fundamental shift – from density-driven scaling to efficiency-driven innovation. Performance is increasingly constrained by power limits and interconnect architecture, while design complexity and verification challenges continue to grow. The surge in compute demand – particularly for AI – has intensified the need for energy-efficient architectures and optimised data movement. This talk explores how design approaches are evolving, from off-the-shelf IP and early SoCs to domain-specific accelerators, advanced packaging, and chiplet-based systems. It highlights the transition toward holistic co-design across software architecture, microarchitecture, and physical design to sustain performance, scalability, and energy efficiency in next-generation silicon.

Biography

Mrudula is a silicon design leader with deep expertise in datapath architecture and implementation. Over a 27-year career, spanning systems designed for 120 MHz to 2.4 GHz,, she has contributed to the design of a broad range of systems, including DSPs, CPUs, GPUs, AI processors, and complex SoCs. Her work has enabled applications across diverse domains, from early USB and WiFi connectivity to mobile computing, digital home platforms, and large-scale AI training and inference systems. She is an inventor on 17 patents. Mru currently leads the GPU design team at Imagination Technologies, focusing on high-performance, energy-efficient compute architectures.