Research Areas

Advancing the science and engineering of future computing systems through innovations spanning algorithm design, computer architecture, circuits, packaging technologies, and machine learning.

Research Area 1

AI Accelerators

As AI models continue to grow in complexity and scale, general-purpose computing platforms struggle to meet performance and energy-efficiency demands.

CompSi develops specialized accelerator architectures optimized for deep learning, foundation models, sparse computation, and emerging AI workloads. Our work focuses on maximizing throughput while minimizing latency, power consumption, and resource utilization.

Research Area 2

Compute-in-Memory Systems

Data movement dominates the energy cost of modern computing. Traditional architectures separate processing and memory, creating significant performance bottlenecks.

Our compute-in-memory research explores architectures that perform computation directly within memory structures, enabling dramatically improved efficiency for AI inference and data-intensive applications.

Research Area 3

Chiplet-Based Architectures

Future computing systems require flexible and scalable integration strategies.

CompSi investigates heterogeneous chiplet technologies, advanced packaging approaches, and high-performance interconnects that enable composable computing platforms. These systems provide an efficient path toward future AI infrastructure and high-performance computing platforms.

Research Area 4

Hardware-Software Co-Design

The highest-performing systems emerge when algorithms and hardware are designed together.

Our research spans programming frameworks, compiler optimization, architecture exploration, and silicon implementation to ensure AI applications fully leverage emerging hardware capabilities.